The aim of this prospective single-institution clinical trial (NCT02002455) was to evaluate the p... more The aim of this prospective single-institution clinical trial (NCT02002455) was to evaluate the potential of advanced post-processing methods for 18 f-fluciclovine pet and multisequence multiparametric MRi in the prediction of prostate cancer (PCa) aggressiveness, defined by Gleason Grade Group (GGG). 21 patients with PCa underwent PET/CT, PET/MRI and MRI before prostatectomy. DWI was post-processed using kurtosis (ADC k , K), mono-(ADC m), and biexponential functions (f, D p , D f) while Logan plots were used to calculate volume of distribution (V t). In total, 16 unique PET (V t , SUV) and MRI derived quantitative parameters were evaluated. Univariate and multivariate analysis were carried out to estimate the potential of the quantitative parameters and their combinations to predict GGG 1 vs >1, using logistic regression with a nested leave-pair out cross validation (LPOCV) scheme and recursive feature elimination technique applied for feature selection. The second order rotating frame imaging (RAFF), monoexponential and kurtosis derived parameters had LPOCV AUC in the range of 0.72 to 0.92 while the corresponding value for V t was 0.85. t he best performance for GGG prediction was achieved by K parameter of kurtosis function followed by quantitative parameters based on DWI, RAFF and 18 F-FACBC PET. No major improvement was achieved using parameter combinations with or without feature selection. Addition of 18 F-FACBC PET derived parameters (V t , SUV) to DWI and RAFF derived parameters did not improve LPOCV AUC. Prostate cancer (PCa) has wide range of aggressiveness, ranging from indolent disease to highly aggressiveness PCa 1,2. Approximately 10-15% of men who undergo radical treatment (surgery or radiotherapy) for localized PCa will develop recurrence based on elevated blood levels of prostate-specific antigen (PSA) 3. Radical treatment (surgery or radiotherapy) has side effects such as impotence and urinary incontinency 4,5. Thus, accurate risk stratification of PCa is of utmost importance to both improve quality of life and PCa-specific survival. Prostate cancer
We aimed to develop deep machine learning (DL) models to improve the detection and segmentation o... more We aimed to develop deep machine learning (DL) models to improve the detection and segmentation of intraprostatic lesions (IL) on bp-MRI by using whole amount prostatectomy specimen-based delineations. We also aimed to investigate whether transfer learning and self-training would improve results with small amount labelled data. Methods: 158 patients had suspicious lesions delineated on MRI based on bp-MRI, 64 patients had ILs delineated on MRI based on whole mount prostatectomy specimen sections, 40 patients were unlabelled. A non-local Mask R-CNN was proposed to improve the segmentation accuracy. Transfer learning was investigated by fine-tuning a model trained using MRI-based delineations with prostatectomy-based delineations. Two label selection strategies were investigated in self-training. The performance of models was evaluated by 3D detection rate, dice similarity coefficient (DSC), 95 percentile Hausdrauff (95 HD, mm) and true positive ratio (TPR). Results: With prostatectomy-based delineations, the non-local Mask R-CNN with fine-tuning and self-training significantly improved all evaluation metrics. For the model with the highest detection rate and DSC, 80.5% (33/41) of lesions in all Gleason Grade Groups (GGG) were detected with DSC of 0.548±0.165, 95 HD of 5.72±3.17 and TPR of 0.613±0.193. Among them, 94.7% (18/19) of lesions with GGG > 2 were detected with DSC of 0.604±0.135, 95 HD of 6.26±3.44 and TPR of 0.580±0.190. Conclusion: DL models can achieve high prostate cancer detection and segmentation accuracy on bp-MRI based on annotations from histologic images. To further improve the performance, more data with annotations of both MRI and whole amount prostatectomy specimens are required.
INTRODUCTION AND OBJECTIVE: To develop and validate Gleason score classifier using texture featur... more INTRODUCTION AND OBJECTIVE: To develop and validate Gleason score classifier using texture features and advance machine learning algorithms of rapid prostate T2-weighted magnetic resonance imaging (T2w) with acquisition time less than 2 minutes. METHODS: Eighty-two prospectively enrolled patients with histologically confirmed prostate cancer (PCa) underwent 3 Tesla magnetic resonance imaging before prostatectomy. T2-weighted magnetic resonance imaging (T2w) was performed using single shot turbo spin echo sequence with repetition time/echo time 4668/ 130 ms, field of view 250x250 mm2, acquisition matrix size 250x320, and acquisition time 1 minute 10 seconds. A histogram alignment method was used to correct a non-standardness of T2w (“intensity drift”). Prostate cancer lesions were delineated using whole mount prostatectomy sections as the ground true. In total, 1631 unique texture features were extracted including Gabor function, Haar transform, image moments, Sobel operator, local binary patterns (LBP), gray-level co-occurrence matrix (GLCM). Classifier was built using logistic regression with either L1 or L2 regularization to compensate the high dimensionality of the data by penalizing large coefficient values of the inferred linear models. The classification performance (Gleason score 3þ3 vs >3þ3), was evaluated by area under a receiver operating characteristic curve (AUC) values. The classification performance of the model built by the regularized logistic regression algorithms was estimated by a nested cross validation strategy with an outer leave-pair-out cross-validation and an inner 10-fold cross validation for hyperparameter selection (Figure 1). RESULTS: The final data set was composed of 126 PCa lesions, 36 and 90 lesions had Gleason score 3þ3 and >3þ3, respectively. The best performing texture features belonged to GLCM and Gabor function groups. The classifier achieved AUC (95% confidence interval) of 0.85 (0.74 0.92). CONCLUSIONS: Machine learning classifier using radiomic and texture features of 2-minute prostate MRI demonstrated a good performance in the classification of prostate cancer Gleason score. Rapid T2-weighted imaging with acquisition time less than 2 minutes and advanced machine learning are promising tools for non-invasive Gleason score prediction.
We evaluated the potential of Relaxation Along a Fictitious Field in second rotating frame (TRAFF... more We evaluated the potential of Relaxation Along a Fictitious Field in second rotating frame (TRAFF2), continues wave T1rho (T1ρcw), adiabatic T1rho (T1ρadiab), adiabatic T2rho (T2ρadiab), DWI and anatomical MRI to differentiate isohydrogenase dehydrogenase (IDH) status between IDH wild type and IDH mutation in 22 patients with glioma. In voxel level analysis, DWI derived parameters using bi-exponential model (0-4000 s/mm2) provided improved performance in classification of IDH status of brain gliomas compared with TRAFF2, T1ρcw, T1ρadiab, T2ρadiab and anatomical MRI.
Max Peters*, Utrecht, Netherlands; David Eldred-Evans, London, United Kingdom; Piet Kurver, Utrec... more Max Peters*, Utrecht, Netherlands; David Eldred-Evans, London, United Kingdom; Piet Kurver, Utrecht, Netherlands; Ugo Giovanni Falagario, Foggia, Italy; Martin J. Connor, London, United Kingdom; Joost J.C. Verhoeff, Utrecht, Netherlands; Giuseppe Carrieri, Luigi Cormio, Foggia, Italy; Pekka Taimen, Hannu J Aronen, Juha Knaapila, Ileana Montoya Perez, Otto Ettala, Turku, Finland; Armando Stabile, Giorgio Gandaglia, Nicola Fossati, Alberto Martini, Vito Cucchiara, Alberto Briganti, Milan, Italy; Anna Lantz, Solna, Sweden; Wolfgang Picker, Oslo, Norway; Erik Haug, Tønsberg, Norway; Tobias Nordstr€ om, Stockholm, Sweden; Mariana Bertoncelli Tanaka, Feargus Hosking-Jervis, Deepika Reddy, Edward Bass, London, United Kingdom; Peter S.N. van Rossum, Utrecht, Netherlands; Suchita Joshi, Elizabeth Pegers, Kathie Wong, Henry Tam, David Hrouda, London, United Kingdom; Stuart McCraken, Sunderland, United Kingdom; Mathias Winkler, Stephen Gordon, Hasan Qazi, London, United Kingdom; Peter J. Bostr€ om, Ivan Jambor, Turku, Finland; Hashim U. Ahmed, London, United Kingdom
ABSTRACTPurposeAlthough prostate cancer is the most common cancer in men in Western countries, th... more ABSTRACTPurposeAlthough prostate cancer is the most common cancer in men in Western countries, there is significant variability in geographical incidence. This might result from genetic factors, discrepancies in screening policies or differences in lifestyle. Gut microbiota has been recently associated with cancer progression, but its role in prostate cancer is unclear.MethodsIn a prospective multicenter clinical trial (NCT02241122), the gut microbiota profiles of 181 men with a clinical suspicion of prostate cancer were assessed utilizing 16S rRNA gene sequencing. Sequences were assigned to operational taxonomic units, and differential abundance analysis, α- and β-diversities, and predictive functional (PICRUSt) analyses were performed. Additionally, plasma steroid hormone levels were correlated with the predicted microbiota functions.ResultsSeveral differences in the gut microbiota between the subjects with and without prostate cancer were noted. Prevotella 9, members of the Erysi...
patients (15%) harboured lymph node metastases on pathological evaluation (N1). The sensitivity, ... more patients (15%) harboured lymph node metastases on pathological evaluation (N1). The sensitivity, specificity, positive predictive value and negative predictive value of F-PSMA PET/CT for N1 was 41% (CI 19%-66%), 94% (CI 86%-97%), 54% (CI 26%-79%) and 90% (CI 81%94%), respectively. Interobserver agreement on imaging N-stage was 95% (CI 90-98%). Additionally, the detection of locally advanced tumour growth (T-stage 3-4) was evaluated, for which F-PSMA PET/CT had a sensitivity, specificity, PPV and NPV of 44% (CI 32-57%), 94% (CI 83%98%), 90% (CI 72%-97%) and 59% (CI 47-69%). CONCLUSIONS: F-PSMA PET/CT has a high specificity (94%), yet a limited sensitivity (41%) for the detection of pelvic lymph node metastases in patients with primary PCa. Promising specificity for detection of locally-advanced tumour stages with F-PSMA PET/CT was observed.
INTRODUCTION AND OBJECTIVE:Bi-parametric MRI (bpMRI: T2W MRI and Apparent Diffusion Coefficient m... more INTRODUCTION AND OBJECTIVE:Bi-parametric MRI (bpMRI: T2W MRI and Apparent Diffusion Coefficient maps (ADC) derived from diffusion weighted imaging) is increasingly being used to characterize prosta...
Background: Accurate detection of clinically significant prostate cancer (csPCa), Gleason Grade G... more Background: Accurate detection of clinically significant prostate cancer (csPCa), Gleason Grade Group ≥ 2, remains a challenge. Prostate MRI radiomics and blood kallikreins have been proposed as tools to improve the performance of biparametric MRI (bpMRI). Purpose: To develop and validate radiomics and kallikrein models for the detection of csPCa. Study Type: Retrospective. Population: A total of 543 men with a clinical suspicion of csPCa, 411 (76%, 411/543) had kallikreins available and 360 (88%, 360/411) did not take 5-alpha-reductase inhibitors. Two data splits into training, validation (split 1: single center, n = 72; split 2: random 50% of pooled datasets from all four centers), and testing (split 1: 4 centers, n = 288; split 2: remaining 50%) were evaluated. Field strength/Sequence: A 3 T/1.5 T, TSE T2-weighted imaging, 3x SE DWI. Assessment: In total, 20,363 radiomic features calculated from manually delineated whole gland (WG) and bpMRI suspicion lesion masks were evaluated in addition to clinical parameters, prostate-specific antigen, four kallikreins, MRI-based qualitative (PI-RADSv2.1/IMPROD bpMRI Likert) scores.
BackgroundIn preclinical models of multiple sclerosis (MS), both adiabatic T1rho (T1ρadiab) and r... more BackgroundIn preclinical models of multiple sclerosis (MS), both adiabatic T1rho (T1ρadiab) and relaxation along a fictitious field (RAFF) imaging have demonstrated potential to noninvasively characterize MS.PurposeTo evaluate the feasibility of whole brain T1ρadiab and RAFF imaging in healthy volunteers and patients with MS.Study TypeSingle institutional clinical trial.Subjects38 healthy volunteers (24–69 years) and 21 patients (26–59 years) with MS. Five healthy volunteers underwent a second MR examination performed within 8 days. Clinical disease severity (The Expanded Disability Status Scale [EDSS] and The Multiple Sclerosis Severity Score [MSSS]) was evaluated at baseline and 1‐year follow‐up (FU).Field Strength/SequenceRAFF in second rotating frame of reference (RAFF2) was performed at 3 T using 3D‐fast‐field echo with magnetization preparation, RF amplitude of 11.74 μT while the corresponding value for T1ρadiab was 13.50 μT. T1‐, T2‐, and FLAIR‐weighted images were acquired w...
PurposeTo evaluate fitting quality and repeatability of four mathematical models for diffusion we... more PurposeTo evaluate fitting quality and repeatability of four mathematical models for diffusion weighted imaging (DWI) during tumor progression in mouse xenograft model of prostate cancer.MethodsHuman prostate cancer cells (PC-3) were implanted subcutaneously in right hind limbs of 11 immunodeficient mice. Tumor growth was followed by weekly DWI examinations using a 7T MR scanner. Additional DWI examination was performed after repositioning following the fourth DWI examination to evaluate short term repeatability. DWI was performed using 15 and 12 b-values in the ranges of 0-500 and 0-2000 s/mm2, respectively. Corrected Akaike information criteria and F-ratio were used to evaluate fitting quality of each model (mono-exponential, stretched exponential, kurtosis, and bi-exponential).ResultsSignificant changes were observed in DWI data during the tumor growth, indicated by ADCm, ADCs, and ADCk. Similar results were obtained using low as well as high b-values. No marked changes in model ...
MRI is a common method of prostate cancer diagnosis. Several MRI‐derived markers, including the a... more MRI is a common method of prostate cancer diagnosis. Several MRI‐derived markers, including the apparent diffusion coefficient (ADC) based on diffusion‐weighted imaging, have been shown to provide values for prostate cancer detection and characterization. The hypothesis of the study was that docetaxel chemotherapy response could be picked up earlier with rotating frame relaxation times TRAFF2 and TRAFF4 than with the continuous wave T1ρ, adiabatic T1ρ, adiabatic T2ρ, T1, T2 or water ADC. Human PC3 prostate cancer cells expressing a red fluorescent protein were implanted in 21 male mice. Docetaxel chemotherapy was given once a week starting 1 week after cell implantation for 10 randomly selected mice, while the rest served as a control group (n = 11). The MRI consisted of relaxation along a fictitious field (RAFF) in the second (RAFF2) and fourth (RAFF4) rotating frames, T1 and T2, continuous wave T1ρ, adiabatic T1ρ and adiabatic T2ρ relaxation time measurements and water ADC. MRI wa...
International Journal of Radiation Oncology*Biology*Physics, 2021
PURPOSE/OBJECTIVE(S) We aim to develop deep learning (DL) models to accurately detect and segment... more PURPOSE/OBJECTIVE(S) We aim to develop deep learning (DL) models to accurately detect and segment intraprostatic lesions (IL) on biparametric MRI (bp-MRI). MATERIALS/METHODS Three patient cohorts with ground truth IL delineated on different modalities were collected. 158 patients from two datasets had suspicious ILs delineated based on bp-MRI: 97 patients were from PROSTATEx-2 Challenge with biopsy result independent from bp-MRI based delineation, 61 patients were from IMPROD clinical Trial with biopsy done for each delineation; 64 patients from IMPROD clinical Trial had ILs identified and delineated by using whole mount prostatectomy specimen sections as reference standard; 40 private patients were unlabeled. We proposed a non-local Mask R-CNN to improve segmentation accuracy by addressing the imperfect registration issue between MRI modalities. We also proposed to post aggregate 2D predictions to estimate IL volumes within the whole prostatic gland and keep top-5 3D predictions for each patient. In order to explore the small dataset problem, we employed different learning techniques including transfer learning and semi-supervised learning with pseudo labelling. We experimented with two label selection strategies to see how they affected model performance. The first strategy kept only one prediction by referring to biopsy result, in order to minimize false positives; while the second strategy kept all top-5 predictions. 3D top-5 detection rate, dice similarity coefficient (DSC), 95 percentile Hausdorff Distance (95 HD, mm) and true positive ratio (TPR) were our evaluation metrics. We compared DL model prediction with prostatectomy-based ground truth delineation to accurately evaluate the boundary and malignancy level. We separately evaluated ILs of all Gleason Grade Group (GGG) and clinically significant ILs (GGG > 2). RESULTS Main results are summarized in Table 1. CONCLUSION Our proposed method demonstrates state-of-art performance in the detection and segmentation of ILs and shows great effectiveness for clinically significant ILs.
European Journal of Nuclear Medicine and Molecular Imaging, 2021
Purpose To prospectively compare 18F-prostate-specific membrane antigen (PSMA)-1007 positron emis... more Purpose To prospectively compare 18F-prostate-specific membrane antigen (PSMA)-1007 positron emission tomography (PET)/CT, whole-body magnetic resonance imaging (WBMRI) including diffusion-weighted imaging (DWI) and standard computed tomography (CT), in primary nodal staging of prostate cancer (PCa). Methods Men with newly diagnosed unfavourable intermediate- or high-risk PCa prospectively underwent 18F-PSMA-1007 PET/CT, WBMRI with DWI and contrast-enhanced CT within a median of 8 days. Six readers (two for each modality) independently reported pelvic lymph nodes as malignant, equivocal or benign while blinded to the other imaging modalities. Sensitivity, specificity and accuracy were reported according to optimistic (equivocal lesions interpreted as benign) and pessimistic (equivocal lesions interpreted as malignant) analyses. The reference standard diagnosis was based on multidisciplinary consensus meetings where available histopathology, clinical and follow-up data were used. Res...
Existing tools for post-radical prostatectomy (RP) prostate cancer biochemical recurrence (BCR) p... more Existing tools for post-radical prostatectomy (RP) prostate cancer biochemical recurrence (BCR) prognosis rely on human pathologist-derived parameters such as tumor grade, with the resulting inter-reviewer variability. Genomic companion diagnostic tests such as Decipher tend to be tissue destructive, expensive, and not routinely available in most centers. We present a tissue non-destructive method for automated BCR prognosis, termed "Histotyping", that employs computational image analysis of morphologic patterns of prostate tissue from a single, routinely acquired hematoxylin and eosin slide. Patients from two institutions (n = 214) were used to train Histotyping for identifying high-risk patients based on six features of glandular morphology extracted from RP specimens. Histotyping was validated for post-RP BCR prognosis on a separate set of n = 675 patients from five institutions and compared against Decipher on n = 167 patients. Histotyping was prognostic of BCR in the ...
PurposeWe aimed to develop and externally validate a nomogram based on MRI volumetric parameters ... more PurposeWe aimed to develop and externally validate a nomogram based on MRI volumetric parameters and clinical information for deciding when SBx should be performed in addition to TBx in man with suspicious prostate MRI.Materials and methodsRetrospective analyses of single (IMPROD, NCT01864135) and multi-institution (MULTI-IMPROD, NCT02241122) clinical trials. All men underwent a unique rapid biparametric magnetic resonance imaging (IMPROD bpMRI) consisting of T2-weighted imaging and three separate DWI acquisitions. Men with IMPROD bpMRI Likert scores of 3–5 were included. Logistic regression models were developed using IMPROD trial (n = 122) and validated using MULTI-IMPROD trial (n = 262) data. The model’s performance was evaluated in the terms of PCa detection with Gleason Grade Group 1 (clinically insignificant prostate cancer, iPCa) and > 1 (clinically significant prostate cancer, csPCa). Net benefits and decision curve analyses (DCA) were compared. Combined biopsies were use...
Background: Computed tomography (CT) and bone scintigraphy (BS) are the imaging modalities curren... more Background: Computed tomography (CT) and bone scintigraphy (BS) are the imaging modalities currently used for distant metastasis staging of prostate cancer (PCa). Objective: To compare standard staging modalities with newer and potentially more accurate imaging modalities. Design, setting, and participants: This prospective, single-centre trial (NCT03537391) enrolled 80 patients with newly diagnosed high-risk PCa (International Society of Urological Pathology grade group 3 and/or prostate-specific antigen [PSA] 20 and/or cT T3; March 2018-June 2019) to undergo primary metastasis staging with two standard and three advanced imaging modalities. Outcome measurements and statistical analysis: The participants underwent the following five imaging examinations within 2 wk of enrolment and without a
The aim of this prospective single-institution clinical trial (NCT02002455) was to evaluate the p... more The aim of this prospective single-institution clinical trial (NCT02002455) was to evaluate the potential of advanced post-processing methods for 18 f-fluciclovine pet and multisequence multiparametric MRi in the prediction of prostate cancer (PCa) aggressiveness, defined by Gleason Grade Group (GGG). 21 patients with PCa underwent PET/CT, PET/MRI and MRI before prostatectomy. DWI was post-processed using kurtosis (ADC k , K), mono-(ADC m), and biexponential functions (f, D p , D f) while Logan plots were used to calculate volume of distribution (V t). In total, 16 unique PET (V t , SUV) and MRI derived quantitative parameters were evaluated. Univariate and multivariate analysis were carried out to estimate the potential of the quantitative parameters and their combinations to predict GGG 1 vs >1, using logistic regression with a nested leave-pair out cross validation (LPOCV) scheme and recursive feature elimination technique applied for feature selection. The second order rotating frame imaging (RAFF), monoexponential and kurtosis derived parameters had LPOCV AUC in the range of 0.72 to 0.92 while the corresponding value for V t was 0.85. t he best performance for GGG prediction was achieved by K parameter of kurtosis function followed by quantitative parameters based on DWI, RAFF and 18 F-FACBC PET. No major improvement was achieved using parameter combinations with or without feature selection. Addition of 18 F-FACBC PET derived parameters (V t , SUV) to DWI and RAFF derived parameters did not improve LPOCV AUC. Prostate cancer (PCa) has wide range of aggressiveness, ranging from indolent disease to highly aggressiveness PCa 1,2. Approximately 10-15% of men who undergo radical treatment (surgery or radiotherapy) for localized PCa will develop recurrence based on elevated blood levels of prostate-specific antigen (PSA) 3. Radical treatment (surgery or radiotherapy) has side effects such as impotence and urinary incontinency 4,5. Thus, accurate risk stratification of PCa is of utmost importance to both improve quality of life and PCa-specific survival. Prostate cancer
We aimed to develop deep machine learning (DL) models to improve the detection and segmentation o... more We aimed to develop deep machine learning (DL) models to improve the detection and segmentation of intraprostatic lesions (IL) on bp-MRI by using whole amount prostatectomy specimen-based delineations. We also aimed to investigate whether transfer learning and self-training would improve results with small amount labelled data. Methods: 158 patients had suspicious lesions delineated on MRI based on bp-MRI, 64 patients had ILs delineated on MRI based on whole mount prostatectomy specimen sections, 40 patients were unlabelled. A non-local Mask R-CNN was proposed to improve the segmentation accuracy. Transfer learning was investigated by fine-tuning a model trained using MRI-based delineations with prostatectomy-based delineations. Two label selection strategies were investigated in self-training. The performance of models was evaluated by 3D detection rate, dice similarity coefficient (DSC), 95 percentile Hausdrauff (95 HD, mm) and true positive ratio (TPR). Results: With prostatectomy-based delineations, the non-local Mask R-CNN with fine-tuning and self-training significantly improved all evaluation metrics. For the model with the highest detection rate and DSC, 80.5% (33/41) of lesions in all Gleason Grade Groups (GGG) were detected with DSC of 0.548±0.165, 95 HD of 5.72±3.17 and TPR of 0.613±0.193. Among them, 94.7% (18/19) of lesions with GGG > 2 were detected with DSC of 0.604±0.135, 95 HD of 6.26±3.44 and TPR of 0.580±0.190. Conclusion: DL models can achieve high prostate cancer detection and segmentation accuracy on bp-MRI based on annotations from histologic images. To further improve the performance, more data with annotations of both MRI and whole amount prostatectomy specimens are required.
INTRODUCTION AND OBJECTIVE: To develop and validate Gleason score classifier using texture featur... more INTRODUCTION AND OBJECTIVE: To develop and validate Gleason score classifier using texture features and advance machine learning algorithms of rapid prostate T2-weighted magnetic resonance imaging (T2w) with acquisition time less than 2 minutes. METHODS: Eighty-two prospectively enrolled patients with histologically confirmed prostate cancer (PCa) underwent 3 Tesla magnetic resonance imaging before prostatectomy. T2-weighted magnetic resonance imaging (T2w) was performed using single shot turbo spin echo sequence with repetition time/echo time 4668/ 130 ms, field of view 250x250 mm2, acquisition matrix size 250x320, and acquisition time 1 minute 10 seconds. A histogram alignment method was used to correct a non-standardness of T2w (“intensity drift”). Prostate cancer lesions were delineated using whole mount prostatectomy sections as the ground true. In total, 1631 unique texture features were extracted including Gabor function, Haar transform, image moments, Sobel operator, local binary patterns (LBP), gray-level co-occurrence matrix (GLCM). Classifier was built using logistic regression with either L1 or L2 regularization to compensate the high dimensionality of the data by penalizing large coefficient values of the inferred linear models. The classification performance (Gleason score 3þ3 vs >3þ3), was evaluated by area under a receiver operating characteristic curve (AUC) values. The classification performance of the model built by the regularized logistic regression algorithms was estimated by a nested cross validation strategy with an outer leave-pair-out cross-validation and an inner 10-fold cross validation for hyperparameter selection (Figure 1). RESULTS: The final data set was composed of 126 PCa lesions, 36 and 90 lesions had Gleason score 3þ3 and >3þ3, respectively. The best performing texture features belonged to GLCM and Gabor function groups. The classifier achieved AUC (95% confidence interval) of 0.85 (0.74 0.92). CONCLUSIONS: Machine learning classifier using radiomic and texture features of 2-minute prostate MRI demonstrated a good performance in the classification of prostate cancer Gleason score. Rapid T2-weighted imaging with acquisition time less than 2 minutes and advanced machine learning are promising tools for non-invasive Gleason score prediction.
We evaluated the potential of Relaxation Along a Fictitious Field in second rotating frame (TRAFF... more We evaluated the potential of Relaxation Along a Fictitious Field in second rotating frame (TRAFF2), continues wave T1rho (T1ρcw), adiabatic T1rho (T1ρadiab), adiabatic T2rho (T2ρadiab), DWI and anatomical MRI to differentiate isohydrogenase dehydrogenase (IDH) status between IDH wild type and IDH mutation in 22 patients with glioma. In voxel level analysis, DWI derived parameters using bi-exponential model (0-4000 s/mm2) provided improved performance in classification of IDH status of brain gliomas compared with TRAFF2, T1ρcw, T1ρadiab, T2ρadiab and anatomical MRI.
Max Peters*, Utrecht, Netherlands; David Eldred-Evans, London, United Kingdom; Piet Kurver, Utrec... more Max Peters*, Utrecht, Netherlands; David Eldred-Evans, London, United Kingdom; Piet Kurver, Utrecht, Netherlands; Ugo Giovanni Falagario, Foggia, Italy; Martin J. Connor, London, United Kingdom; Joost J.C. Verhoeff, Utrecht, Netherlands; Giuseppe Carrieri, Luigi Cormio, Foggia, Italy; Pekka Taimen, Hannu J Aronen, Juha Knaapila, Ileana Montoya Perez, Otto Ettala, Turku, Finland; Armando Stabile, Giorgio Gandaglia, Nicola Fossati, Alberto Martini, Vito Cucchiara, Alberto Briganti, Milan, Italy; Anna Lantz, Solna, Sweden; Wolfgang Picker, Oslo, Norway; Erik Haug, Tønsberg, Norway; Tobias Nordstr€ om, Stockholm, Sweden; Mariana Bertoncelli Tanaka, Feargus Hosking-Jervis, Deepika Reddy, Edward Bass, London, United Kingdom; Peter S.N. van Rossum, Utrecht, Netherlands; Suchita Joshi, Elizabeth Pegers, Kathie Wong, Henry Tam, David Hrouda, London, United Kingdom; Stuart McCraken, Sunderland, United Kingdom; Mathias Winkler, Stephen Gordon, Hasan Qazi, London, United Kingdom; Peter J. Bostr€ om, Ivan Jambor, Turku, Finland; Hashim U. Ahmed, London, United Kingdom
ABSTRACTPurposeAlthough prostate cancer is the most common cancer in men in Western countries, th... more ABSTRACTPurposeAlthough prostate cancer is the most common cancer in men in Western countries, there is significant variability in geographical incidence. This might result from genetic factors, discrepancies in screening policies or differences in lifestyle. Gut microbiota has been recently associated with cancer progression, but its role in prostate cancer is unclear.MethodsIn a prospective multicenter clinical trial (NCT02241122), the gut microbiota profiles of 181 men with a clinical suspicion of prostate cancer were assessed utilizing 16S rRNA gene sequencing. Sequences were assigned to operational taxonomic units, and differential abundance analysis, α- and β-diversities, and predictive functional (PICRUSt) analyses were performed. Additionally, plasma steroid hormone levels were correlated with the predicted microbiota functions.ResultsSeveral differences in the gut microbiota between the subjects with and without prostate cancer were noted. Prevotella 9, members of the Erysi...
patients (15%) harboured lymph node metastases on pathological evaluation (N1). The sensitivity, ... more patients (15%) harboured lymph node metastases on pathological evaluation (N1). The sensitivity, specificity, positive predictive value and negative predictive value of F-PSMA PET/CT for N1 was 41% (CI 19%-66%), 94% (CI 86%-97%), 54% (CI 26%-79%) and 90% (CI 81%94%), respectively. Interobserver agreement on imaging N-stage was 95% (CI 90-98%). Additionally, the detection of locally advanced tumour growth (T-stage 3-4) was evaluated, for which F-PSMA PET/CT had a sensitivity, specificity, PPV and NPV of 44% (CI 32-57%), 94% (CI 83%98%), 90% (CI 72%-97%) and 59% (CI 47-69%). CONCLUSIONS: F-PSMA PET/CT has a high specificity (94%), yet a limited sensitivity (41%) for the detection of pelvic lymph node metastases in patients with primary PCa. Promising specificity for detection of locally-advanced tumour stages with F-PSMA PET/CT was observed.
INTRODUCTION AND OBJECTIVE:Bi-parametric MRI (bpMRI: T2W MRI and Apparent Diffusion Coefficient m... more INTRODUCTION AND OBJECTIVE:Bi-parametric MRI (bpMRI: T2W MRI and Apparent Diffusion Coefficient maps (ADC) derived from diffusion weighted imaging) is increasingly being used to characterize prosta...
Background: Accurate detection of clinically significant prostate cancer (csPCa), Gleason Grade G... more Background: Accurate detection of clinically significant prostate cancer (csPCa), Gleason Grade Group ≥ 2, remains a challenge. Prostate MRI radiomics and blood kallikreins have been proposed as tools to improve the performance of biparametric MRI (bpMRI). Purpose: To develop and validate radiomics and kallikrein models for the detection of csPCa. Study Type: Retrospective. Population: A total of 543 men with a clinical suspicion of csPCa, 411 (76%, 411/543) had kallikreins available and 360 (88%, 360/411) did not take 5-alpha-reductase inhibitors. Two data splits into training, validation (split 1: single center, n = 72; split 2: random 50% of pooled datasets from all four centers), and testing (split 1: 4 centers, n = 288; split 2: remaining 50%) were evaluated. Field strength/Sequence: A 3 T/1.5 T, TSE T2-weighted imaging, 3x SE DWI. Assessment: In total, 20,363 radiomic features calculated from manually delineated whole gland (WG) and bpMRI suspicion lesion masks were evaluated in addition to clinical parameters, prostate-specific antigen, four kallikreins, MRI-based qualitative (PI-RADSv2.1/IMPROD bpMRI Likert) scores.
BackgroundIn preclinical models of multiple sclerosis (MS), both adiabatic T1rho (T1ρadiab) and r... more BackgroundIn preclinical models of multiple sclerosis (MS), both adiabatic T1rho (T1ρadiab) and relaxation along a fictitious field (RAFF) imaging have demonstrated potential to noninvasively characterize MS.PurposeTo evaluate the feasibility of whole brain T1ρadiab and RAFF imaging in healthy volunteers and patients with MS.Study TypeSingle institutional clinical trial.Subjects38 healthy volunteers (24–69 years) and 21 patients (26–59 years) with MS. Five healthy volunteers underwent a second MR examination performed within 8 days. Clinical disease severity (The Expanded Disability Status Scale [EDSS] and The Multiple Sclerosis Severity Score [MSSS]) was evaluated at baseline and 1‐year follow‐up (FU).Field Strength/SequenceRAFF in second rotating frame of reference (RAFF2) was performed at 3 T using 3D‐fast‐field echo with magnetization preparation, RF amplitude of 11.74 μT while the corresponding value for T1ρadiab was 13.50 μT. T1‐, T2‐, and FLAIR‐weighted images were acquired w...
PurposeTo evaluate fitting quality and repeatability of four mathematical models for diffusion we... more PurposeTo evaluate fitting quality and repeatability of four mathematical models for diffusion weighted imaging (DWI) during tumor progression in mouse xenograft model of prostate cancer.MethodsHuman prostate cancer cells (PC-3) were implanted subcutaneously in right hind limbs of 11 immunodeficient mice. Tumor growth was followed by weekly DWI examinations using a 7T MR scanner. Additional DWI examination was performed after repositioning following the fourth DWI examination to evaluate short term repeatability. DWI was performed using 15 and 12 b-values in the ranges of 0-500 and 0-2000 s/mm2, respectively. Corrected Akaike information criteria and F-ratio were used to evaluate fitting quality of each model (mono-exponential, stretched exponential, kurtosis, and bi-exponential).ResultsSignificant changes were observed in DWI data during the tumor growth, indicated by ADCm, ADCs, and ADCk. Similar results were obtained using low as well as high b-values. No marked changes in model ...
MRI is a common method of prostate cancer diagnosis. Several MRI‐derived markers, including the a... more MRI is a common method of prostate cancer diagnosis. Several MRI‐derived markers, including the apparent diffusion coefficient (ADC) based on diffusion‐weighted imaging, have been shown to provide values for prostate cancer detection and characterization. The hypothesis of the study was that docetaxel chemotherapy response could be picked up earlier with rotating frame relaxation times TRAFF2 and TRAFF4 than with the continuous wave T1ρ, adiabatic T1ρ, adiabatic T2ρ, T1, T2 or water ADC. Human PC3 prostate cancer cells expressing a red fluorescent protein were implanted in 21 male mice. Docetaxel chemotherapy was given once a week starting 1 week after cell implantation for 10 randomly selected mice, while the rest served as a control group (n = 11). The MRI consisted of relaxation along a fictitious field (RAFF) in the second (RAFF2) and fourth (RAFF4) rotating frames, T1 and T2, continuous wave T1ρ, adiabatic T1ρ and adiabatic T2ρ relaxation time measurements and water ADC. MRI wa...
International Journal of Radiation Oncology*Biology*Physics, 2021
PURPOSE/OBJECTIVE(S) We aim to develop deep learning (DL) models to accurately detect and segment... more PURPOSE/OBJECTIVE(S) We aim to develop deep learning (DL) models to accurately detect and segment intraprostatic lesions (IL) on biparametric MRI (bp-MRI). MATERIALS/METHODS Three patient cohorts with ground truth IL delineated on different modalities were collected. 158 patients from two datasets had suspicious ILs delineated based on bp-MRI: 97 patients were from PROSTATEx-2 Challenge with biopsy result independent from bp-MRI based delineation, 61 patients were from IMPROD clinical Trial with biopsy done for each delineation; 64 patients from IMPROD clinical Trial had ILs identified and delineated by using whole mount prostatectomy specimen sections as reference standard; 40 private patients were unlabeled. We proposed a non-local Mask R-CNN to improve segmentation accuracy by addressing the imperfect registration issue between MRI modalities. We also proposed to post aggregate 2D predictions to estimate IL volumes within the whole prostatic gland and keep top-5 3D predictions for each patient. In order to explore the small dataset problem, we employed different learning techniques including transfer learning and semi-supervised learning with pseudo labelling. We experimented with two label selection strategies to see how they affected model performance. The first strategy kept only one prediction by referring to biopsy result, in order to minimize false positives; while the second strategy kept all top-5 predictions. 3D top-5 detection rate, dice similarity coefficient (DSC), 95 percentile Hausdorff Distance (95 HD, mm) and true positive ratio (TPR) were our evaluation metrics. We compared DL model prediction with prostatectomy-based ground truth delineation to accurately evaluate the boundary and malignancy level. We separately evaluated ILs of all Gleason Grade Group (GGG) and clinically significant ILs (GGG > 2). RESULTS Main results are summarized in Table 1. CONCLUSION Our proposed method demonstrates state-of-art performance in the detection and segmentation of ILs and shows great effectiveness for clinically significant ILs.
European Journal of Nuclear Medicine and Molecular Imaging, 2021
Purpose To prospectively compare 18F-prostate-specific membrane antigen (PSMA)-1007 positron emis... more Purpose To prospectively compare 18F-prostate-specific membrane antigen (PSMA)-1007 positron emission tomography (PET)/CT, whole-body magnetic resonance imaging (WBMRI) including diffusion-weighted imaging (DWI) and standard computed tomography (CT), in primary nodal staging of prostate cancer (PCa). Methods Men with newly diagnosed unfavourable intermediate- or high-risk PCa prospectively underwent 18F-PSMA-1007 PET/CT, WBMRI with DWI and contrast-enhanced CT within a median of 8 days. Six readers (two for each modality) independently reported pelvic lymph nodes as malignant, equivocal or benign while blinded to the other imaging modalities. Sensitivity, specificity and accuracy were reported according to optimistic (equivocal lesions interpreted as benign) and pessimistic (equivocal lesions interpreted as malignant) analyses. The reference standard diagnosis was based on multidisciplinary consensus meetings where available histopathology, clinical and follow-up data were used. Res...
Existing tools for post-radical prostatectomy (RP) prostate cancer biochemical recurrence (BCR) p... more Existing tools for post-radical prostatectomy (RP) prostate cancer biochemical recurrence (BCR) prognosis rely on human pathologist-derived parameters such as tumor grade, with the resulting inter-reviewer variability. Genomic companion diagnostic tests such as Decipher tend to be tissue destructive, expensive, and not routinely available in most centers. We present a tissue non-destructive method for automated BCR prognosis, termed "Histotyping", that employs computational image analysis of morphologic patterns of prostate tissue from a single, routinely acquired hematoxylin and eosin slide. Patients from two institutions (n = 214) were used to train Histotyping for identifying high-risk patients based on six features of glandular morphology extracted from RP specimens. Histotyping was validated for post-RP BCR prognosis on a separate set of n = 675 patients from five institutions and compared against Decipher on n = 167 patients. Histotyping was prognostic of BCR in the ...
PurposeWe aimed to develop and externally validate a nomogram based on MRI volumetric parameters ... more PurposeWe aimed to develop and externally validate a nomogram based on MRI volumetric parameters and clinical information for deciding when SBx should be performed in addition to TBx in man with suspicious prostate MRI.Materials and methodsRetrospective analyses of single (IMPROD, NCT01864135) and multi-institution (MULTI-IMPROD, NCT02241122) clinical trials. All men underwent a unique rapid biparametric magnetic resonance imaging (IMPROD bpMRI) consisting of T2-weighted imaging and three separate DWI acquisitions. Men with IMPROD bpMRI Likert scores of 3–5 were included. Logistic regression models were developed using IMPROD trial (n = 122) and validated using MULTI-IMPROD trial (n = 262) data. The model’s performance was evaluated in the terms of PCa detection with Gleason Grade Group 1 (clinically insignificant prostate cancer, iPCa) and > 1 (clinically significant prostate cancer, csPCa). Net benefits and decision curve analyses (DCA) were compared. Combined biopsies were use...
Background: Computed tomography (CT) and bone scintigraphy (BS) are the imaging modalities curren... more Background: Computed tomography (CT) and bone scintigraphy (BS) are the imaging modalities currently used for distant metastasis staging of prostate cancer (PCa). Objective: To compare standard staging modalities with newer and potentially more accurate imaging modalities. Design, setting, and participants: This prospective, single-centre trial (NCT03537391) enrolled 80 patients with newly diagnosed high-risk PCa (International Society of Urological Pathology grade group 3 and/or prostate-specific antigen [PSA] 20 and/or cT T3; March 2018-June 2019) to undergo primary metastasis staging with two standard and three advanced imaging modalities. Outcome measurements and statistical analysis: The participants underwent the following five imaging examinations within 2 wk of enrolment and without a
Uploads
Papers by Hannu Aronen