Deep Learning Enables Prostate MRI Segmentation: A Large Cohort Evaluation With Inter-Rater Variability Analysis

Whole-prostate gland (WPG) segmentation plays a significant role in prostate volume measurement, treatment, and biopsy planning. This study evaluated a previously developed automatic WPG segmentation, deep attentive neural network (DANN), on a large, continuous patient cohort to test its feasibility in a clinical setting. With IRB approval and HIPAA compliance, the study cohort included 3,698 3T MRI scans acquired between 2016 and 2020. In total, 335 MRI scans were used to train the model, and 3,210 and 100 were used to conduct the qualitative and quantitative evaluation of the model. In addition, the DANN-enabled prostate volume estimation was evaluated by using 50 MRI scans in comparison with manual prostate volume estimation. For qualitative evaluation, visual grading was used to evaluate the performance of WPG segmentation by two abdominal radiologists, and DANN demonstrated either acceptable or excellent performance in over 96% of the testing cohort on the WPG or each prostate sub-portion (apex, midgland, or base). Two radiologists reached a substantial agreement on WPG and midgland segmentation (κ = 0.75 and 0.63) and moderate agreement on apex and base segmentation (κ = 0.56 and 0.60). For quantitative evaluation, DANN demonstrated a dice similarity coefficient of 0.93 ± 0.02, significantly higher than other baseline methods, such as DeepLab v3+ and UNet (both p values < 0.05). For the volume measurement, 96% of the evaluation cohort achieved differences between the DANN-enabled and manual volume measurement within 95% limits of agreement. In conclusion, the study showed that the DANN achieved sufficient and consistent WPG segmentation on a large, continuous study cohort, demonstrating its great potential to serve as a tool to measure prostate volume.

Download Full-text

Prostate Volume Measurement by Ultrasonography Over 40 Years Age of Bangladeshi Population

Annals of International medical and Dental Research ◽

10.53339/aimdr.2022.8.1.44 ◽

2022 ◽

Vol 8 (1) ◽

pp. 350-356

Author(s):

Towhida Naheen

Keyword(s):

Benign Prostatic Hyperplasia ◽

Prostate Volume ◽

Prostate Gland ◽

Digital Rectal Examination ◽

Volume Measurement ◽

Prostatic Hyperplasia ◽

P Value ◽

College Hospital ◽

Histologic Diagnosis ◽

Bangladeshi Population

Background: Benign prostatic hyperplasia (BPH) or benign prostatic hypertrophy, is a histologic diagnosis status characterized by proliferation of the ‘glandular elements’ of the prostate, which may lead to an enlarged prostate gland. In many studies, people over the age of 40 years found as the most vulnerable for BPH. Ultrasonography is a prominent method to determine prostate volume or size. Aim of the study: The aim of the present study was to evaluate the prostate volume measurement for the Bangladeshi population over the age of 40 years by ultrasonography.Methods:This prospective, observational study was conducted in the Department of Anatomy, Chattogram Medical College Hospital, Chattogram, Bangladesh during the period from January 2019 to December 2020. In total 157 suspected patients of benign prostatic hyperplasia were selected as the study population. All patients were clinically diagnosed for BPH, based on the present prostate symptoms and digital rectal examination. To measure the prostate volume, abdominal ultrasonography was performed for all the patients. After enucleation, another ultrasonogram was performed for all the patients to measure the existing sizes of the prostates of the patients. All the data were processed, analyzed, and disseminated by MS-word and SPSS programs as per need.Results:Finally, in this study in analyzing the volumes of the prostates of the participants according to the abdominal ultra-sonographic reports of pre-operative stage we observed, in 9%, 34%, 31%, 30%, 21% and 32% patients, the prostate sizes (In cc) were <20, 21-40, 41-60, 61-80, 81-100 and >100 cc respectively. On the other hand, after enucleation, in 11.46%, 24.20%, 28.66%, 27.39%, 7.01% and 1.27% patients, the prostate sizes (In cc) were found <20, 21-40, 41-60, 61-80, 81-100 and >100 cc respectively. The mean changes of prostate sizes between pre- and post-operative stages among the participant was not significant where the P value was found 0.464.Conclusion:The findings of this study support the applications of abdominal ultrasonographic evaluation for suspected benign prostatic hyperplasia patients to know about the exact volumes of their prostates for selecting the appropriate surgical approach.

Download Full-text

1014 Assessment of Prostate Gland Volume Estimations Using MRI And TRUS Compared to Radical Prostatectomy Specimens

British Journal of Surgery ◽

10.1093/bjs/znab259.1100 ◽

2021 ◽

Vol 108 (Supplement_6) ◽

Author(s):

S Yallappa ◽

I A Aneke ◽

M Amjad ◽

A Clark ◽

L Gommersall

Keyword(s):

Prostate Cancer ◽

Radical Prostatectomy ◽

Prostate Volume ◽

Prostate Gland ◽

Specific Antigen ◽

Clinical Information ◽

Volume Estimate ◽

Cancer Group ◽

Specimen Volume ◽

Mri Scans

Abstract Introduction The prostate volume is an essential criterion to calculate prostate specific antigen density (PSAD). When selecting patients for active surveillance (AS), in newly diagnosed low risk prostate cancer group or continuing AS in previously diagnosed cancer prostate, PSAD plays a major role. Estimation of the volume using digital rectal exam or PSA are inaccurate. This study aims to conduct a retrospective review to evaluate the accuracy of prostatic volume estimates in patients who had TRUS and MRI scans, comparing the obtained volumes to the reference standard which is the actual volume of radical prostatectomy specimen. Method Data was collected retrospectively for all patients who had robotic assisted radical prostatectomy (RRP) at the Royal Stoke Hospital between October 2015 and October 2018. Clinical information of TRUS and MRI prostate volumes were extracted from PACS and prostate specimen volume was collected from the histopathology report of RRP specimen. Results Pathological specimen prostate volume showed a positive relationship between MRI and TRUS prostate volume with a correlation efficient of 0.71 for MRI vs RRP specimen volume and 0.81 for TRUS vs RRP specimen volume. Mean TRUS volume underestimated prostate volume by 7.33cc and mean MRI volume underestimated prostate volume by 0.02cc Conclusions Although the study showed positive correlation between measuring prostate volume using MRI and TRUS as compared to RRP specimens, MRI showed a greater accuracy as compared to TRUS. We conclude that using MRI prostate volume gives more precise prostate volume estimate aiding appropriate therapeutic planning of patients with prostate cancer.

Download Full-text

Assessment of Artificial Intelligence Automatic Multiple Sclerosis Lesion Delineation Tool for Clinical Use

Clinical Neuroradiology ◽

10.1007/s00062-021-01089-z ◽

2021 ◽

Author(s):

Amalie Monberg Hindsholm ◽

Stig Præstekjær Cramer ◽

Helle Juhl Simonsen ◽

Jette Lautrup Frederiksen ◽

Flemming Andersen ◽

...

Keyword(s):

Multiple Sclerosis ◽

White Matter ◽

Quantitative Evaluation ◽

White Matter Lesions ◽

Qualitative Evaluation ◽

White Matter Hyperintensity ◽

Multiple Sclerosis Lesion ◽

Dice Similarity Coefficient ◽

Accuracy Score ◽

Clinical Value

Abstract Purpose To implement and validate an existing algorithm for automatic delineation of white matter lesions on magnetic resonance imaging (MRI) in patients with multiple sclerosis (MS) on a local single-center dataset. Methods We implemented a white matter hyperintensity segmentation model, based on a 2D convolutional neural network, using the conventional T2-weighted fluid attenuated inversion recovery (FLAIR) MRI sequence as input. The model was adapted for delineation of MS lesions by further training on a local dataset of 93 MS patients with a total of 3040 lesions. A quantitative evaluation was performed on ten test patients, in which model-generated masks were compared to manually delineated masks from two expert delineators. A subsequent qualitative evaluation of the implemented model was performed by two expert delineators, in which generated delineation masks on a clinical dataset of 53 patients were rated acceptable (< 10% errors) or unacceptable (> 10% errors) based on the total number of true lesions. Results The quantitative evaluation resulted in an average accuracy score (F1) of 0.71, recall of 0.77 and dice similarity coefficient of 0.62. Our implemented model obtained the highest scores in all three metrics, when compared to three out of the box lesion segmentation models. In the clinical evaluation an average of 94% of our 53 model-generated masks were rated acceptable. Conclusion After adaptation to our local dataset, the implemented segmentation model was able to delineate MS lesions with a high clinical value as rated by delineation experts while outperforming popular out of the box applications. This serves as a promising step towards implementation of automatic lesion delineation in our MS clinic.

Download Full-text

IMG-22. A DEEP LEARNING MODEL FOR AUTOMATIC POSTERIOR FOSSA PEDIATRIC BRAIN TUMOR SEGMENTATION: A MULTI-INSTITUTIONAL STUDY

Neuro-Oncology ◽

10.1093/neuonc/noaa222.357 ◽

2020 ◽

Vol 22 (Supplement_3) ◽

pp. iii359-iii359

Author(s):

Lydia Tam ◽

Edward Lee ◽

Michelle Han ◽

Jason Wright ◽

Leo Chen ◽

...

Keyword(s):

Deep Learning ◽

Posterior Fossa ◽

Ground Truth ◽

Learning Model ◽

Tumor Segmentation ◽

Dice Similarity Coefficient ◽

Study Cohort ◽

Navigation Systems ◽

Tumor Types ◽

Deep Learning Model

Abstract BACKGROUND Brain tumors are the most common solid malignancies in childhood, many of which develop in the posterior fossa (PF). Manual tumor measurements are frequently required to optimize registration into surgical navigation systems or for surveillance of nonresectable tumors after therapy. With recent advances in artificial intelligence (AI), automated MRI-based tumor segmentation is now feasible without requiring manual measurements. Our goal was to create a deep learning model for automated PF tumor segmentation that can register into navigation systems and provide volume output. METHODS 720 pre-surgical MRI scans from five pediatric centers were divided into training, validation, and testing datasets. The study cohort comprised of four PF tumor types: medulloblastoma, diffuse midline glioma, ependymoma, and brainstem or cerebellar pilocytic astrocytoma. Manual segmentation of the tumors by an attending neuroradiologist served as “ground truth” labels for model training and evaluation. We used 2D Unet, an encoder-decoder convolutional neural network architecture, with a pre-trained ResNet50 encoder. We assessed ventricle segmentation accuracy on a held-out test set using Dice similarity coefficient (0–1) and compared ventricular volume calculation between manual and model-derived segmentations using linear regression. RESULTS Compared to the ground truth expert human segmentation, overall Dice score for model performance accuracy was 0.83 for automatic delineation of the 4 tumor types. CONCLUSIONS In this multi-institutional study, we present a deep learning algorithm that automatically delineates PF tumors and outputs volumetric information. Our results demonstrate applied AI that is clinically applicable, potentially augmenting radiologists, neuro-oncologists, and neurosurgeons for tumor evaluation, surveillance, and surgical planning.

Download Full-text

Prostate volume measurement by transrectal ultrasound and computed tomography before and after permanent prostate brachytherapy

International Journal of Radiation Oncology*Biology*Physics ◽

10.1016/j.ijrobp.2004.04.015 ◽

2004 ◽

Vol 60 (3) ◽

pp. 767-776 ◽

Cited By ~ 19

Author(s):

Matthew C. Solhjem ◽

Brian J. Davis ◽

Thomas M. Pisansky ◽

Torrence M. Wilson ◽

Lance A. Mynderse ◽

...

Keyword(s):

Computed Tomography ◽

Prostate Volume ◽

Transrectal Ultrasound ◽

Volume Measurement ◽

Prostate Brachytherapy ◽

Before And After

Download Full-text

A prospective study of the role of bladder scanning and post-void residual volume measurement in improving diagnostic accuracy of cauda equina syndrome

The Bone & Joint Journal ◽

10.1302/0301-620x.102b6.bjj-2020-0195.r1 ◽

2020 ◽

Vol 102-B (6) ◽

pp. 677-682

Author(s):

Galateia Katzouraki ◽

Akbar Jaleel Zubairi ◽

Oded Hershkovich ◽

Michael P. Grevitt

Keyword(s):

Risk Stratification ◽

Prospective Study ◽

Clinical Assessment ◽

Predictive Value ◽

Cauda Equina Syndrome ◽

Cauda Equina ◽

Cost Savings ◽

Volume Measurement ◽

Cauda Equina Compression ◽

Mri Scans

Aims Diagnosis of cauda equina syndrome (CES) remains difficult; clinical assessment has low accuracy in reliably predicting MRI compression of the cauda equina (CE). This prospective study tests the usefulness of ultrasound bladder scans as an adjunct for diagnosing CES. Methods A total of 260 patients with suspected CES were referred to a tertiary spinal unit over a 16-month period. All were assessed by Board-eligible spinal surgeons and had transabdominal ultrasound bladder scans for pre- and post-voiding residual (PVR) volume measurements before lumbosacral MRI. Results The study confirms the low predictive value of ‘red flag’ symptoms and signs. Of note ‘bilateral sciatica’ had a sensitivity of 32.4%, and a positive predictive value (PPV) of only 17.2%, and negative predictive value (NPV) 88.3%. Use of a PVR volume of ≥ 200 ml was a demonstrably more accurate test for predicting cauda equina compression on subsequent MRI (p < 0.001). The PVR sensitivity was 94.1%, specificity 66.8%, PPV 29.9% and NPV 98.7%. The PVR allowed risk-stratification with 13% patients deemed ‘low-risk’ of CES. They had non-urgent MRI scans. None of the latter scans showed any cauda equina compression (p < 0.006) or individuals developed subsequent CES in the intervening period. There were considerable cost-savings associated with the above strategy. Conclusion This is the largest reported prospective evaluation of suspected CES. Use of the PVR volume ≥ 200 ml was considerably more accurate in predicting CES. It is a useful adjunct to conventional clinical assessment and allows risk-stratification in managing suspected CES. If adopted widely it is less likely incomplete CES would be missed. Cite this article: Bone Joint J 2020;102-B(6):677–682.

Download Full-text

MNG-08 VOLUMETRIC STUDIES IN ASYMPTOMATIC MENINGIOMAS: SLOWDOWN CASES AND GROWTH ARREST CASES

Neuro-Oncology Advances ◽

10.1093/noajnl/vdz039.164 ◽

2019 ◽

Vol 1 (Supplement_2) ◽

pp. ii36-ii36

Author(s):

Ryuichi Hirayama ◽

Tomoyoshi Nakagawa ◽

Toru Umehara ◽

Chisato Yokota ◽

Noriyuki Kijima ◽

...

Keyword(s):

Surgical Treatment ◽

Growth Arrest ◽

Surgical Excision ◽

Volume Measurement ◽

Female Ratio ◽

Volume Increase ◽

Volume Curve ◽

Increase Rate ◽

Mri Scans

Abstract BACKGROUND The opportunity to follow up for asymptomatic meningiomas has increased. We have reported the risk of volume increase by individual continuous volume measurement of asymptomatic meningiomas. However, We have not reached fully understanding about natural history of meningiomas. Among cases are followed up over time, there are some cases that the volume increase rates slows down or almost stops are observed. METHODS We enrolled consecutive adult patients of asymptomatic meningiomas who follow-up for 2 years or more and 3 or more MRI scans. We performed sequential volumetric measurements on 95 patients (105 lesions) who met the criteria. We classified these transient volume curve of each lesion into three groups “Growing”, “Slowdown”, and “Growth arrest” for analysis. RESULTS The average age at the first visit was 62.8 years, the average follow-up period was 61.8 months, and the male-female ratio was 20:75 (male: female). There were 67 cases (73 lesions: 70.9%) that were in increasing trend, and 19 cases of those were received resection. Eleven cases (12 lesions: 11.7%) showed a tendency of “slow down” the increase rate, and one patient who became symptomatic led to surgical excision. In 18 cases (18 lesions: 17.4%) in which almost no volume change was observed during the observation period, no cases resulted in surgical treatment. CONCLUSIONS Among the meningiomas cases that have been followed for a long time, there are not a few those increase rate of tumor volume slows or does not change. Furthermore, most of these cases did not result in surgical treatment. The presence of these “Slowdown” and “Growth arrest” cases at a certain rate may have suggested the possibility of a Gompertz curve model as the natural course of meningiomas.

Download Full-text

How Accurately Can Prostate Gland Imaging Measure the Prostate Gland Volume? Results of a Systematic Review

Prostate Cancer ◽

10.1155/2019/6932572 ◽

2019 ◽

Vol 2019 ◽

pp. 1-12 ◽

Cited By ~ 10

Author(s):

David R. H. Christie ◽

Christopher F. Sharpley

Keyword(s):

Systematic Review ◽

Prostate Gland ◽

Correlation Coefficients ◽

Volume Measurement ◽

Imaging Methods ◽

Gland Volume ◽

Quantitative Measurements ◽

Ultrasound Computed Tomography ◽

Volume Measurements

Aim. The measurement of the volume of the prostate gland can have an influence on many clinical decisions. Various imaging methods have been used to measure it. Our aim was to conduct the first systematic review of their accuracy. Methods. The literature describing the accuracy of imaging methods for measuring the prostate gland volume was systematically reviewed. Articles were included if they compared volume measurements obtained by medical imaging with a reference volume measurement obtained after removal of the gland by radical prostatectomy. Correlation and concordance statistics were summarised. Results. 28 articles describing 7768 patients were identified. The imaging methods were ultrasound, computed tomography, and magnetic resonance imaging (US, CT, and MRI). Wide variations were noted but most articles about US and CT provided correlation coefficients that lay between 0.70 and 0.90, while those describing MRI seemed slightly more accurate at 0.80-0.96. When concordance was reported, it was similar; over- and underestimation of the prostate were variably reported. Most studies showed evidence of at least moderate bias and the quality of the studies was highly variable. Discussion. The reported correlations were moderate to high in strength indicating that imaging is sufficiently accurate when quantitative measurements of prostate gland volume are required. MRI was slightly more accurate than the other methods.

Download Full-text