Training Convolutional Networks for Prostate Segmentation With Limited Data.

Multi-zonal segmentation is a critical component of computer-aided diagnostic systems for detecting and staging prostate cancer. Previously, convolutional neural networks such as the U-Net have been used to produce fully automatic multi-zonal prostate segmentation on magnetic resonance images (MRIs) with performance comparable to human experts, but these often require large amounts of manually segmented training data to produce acceptable results. For institutions that have limited amounts of labeled MRI exams, it is not clear how much data is needed to train a segmentation model, and which training strategy should be used to maximize the value of the available data. This work compares how the strategies of transfer learning and aggregated training using publicly available external data can improve segmentation performance on internal, site-specific prostate MR images, and evaluates how the performance varies with the amount of internal data used for training. Cross training experiments were performed to show that differences between internal and external data were impactful. Using a standard U-Net architecture, optimizations were performed to select between 2D and 3D variants, and to determine the depth of fine-tuning required for optimal transfer learning. With the optimized architecture, the performance of transfer learning and aggregated training were compared for a range of 5-40 internal datasets. The results show that both strategies consistently improve performance and produced segmentation results that are comparable to that of human experts with approximately 20 site-specific MRI datasets. These findings can help guide the development of site-specific prostate segmentation models for both clinical and research applications.

Prostatic peripheral zone thickness: what is normal on magnetic resonance imaging?

PURPOSE: To report the precision of a technique of measuring the PZ thickness on T2-weighted MRI and report normal parameters in patients with normal-sized prostates. We also wanted to establish the mean and second standard deviations (2SD) above and below the mean as criteria for abnormally narrow or expanded PZ thickness. METHODS: Of the initial 1566 consecutive cohort referred for evaluation for carcinoma based on elevated PSA (prostate specific antibody) or DRE (digital rectal examination), 132 separate subjects with normal-sized prostates were selected for this study. Mean age was 58.2 years (15-82). Median serum PSA was 6.2 ng/mL (range 0.3-145). Most were asymptomatic for lower urinary tract symptoms (LUTS). Inclusion criteria in this study required technically adequate T2-weighted MRI and total prostatic volume (TPV) ≤ 25 cc. Exclusion criteria included post-prostatic surgical and radiation patients, patients having had medical management or minimally invasive therapy for BPH, those being treated for prostatitis. Patients with suspected tumor expanding or obscuring measurement boundaries were also not considered. Transition zone (TZ) and peripheral zone (PZ) volumes were determined using the prolate ellipsoid model. Posterolateral measurement of the PZ was obtained at the axial level of maximal transverse diameter of the prostate on a line drawn from the outer boundary of the TZ to the inner boundary of the external prostatic capsule (EPC). The data were normally distributed. Therefore, it was analyzed using the 2-sided student t-test and Pearson product correlation statistic. RESULTS: Mean pooled (composite) measurement for the posterolateral PZ (PLPZ) was 10 mm (CI 9.5-10.5 mm) with SD of 2.87 mm. Means were statistically the same for the 2 observers (p = 0.75). Pearson correlation between the two observers was 0.63. CONCLUSIONS: In a prostate ≤ 25 cc volume , the posterolateral PZ should be no thicker than 15.8 mm and averages 10.0 mm when measured in the maximal axial plane on MRI. These norms were independent of age or use of endorectal coil. The technique measurement demonstrated clinically useful precision.

Measurement of Prostate Volume with MRI (A Guide for the Perplexed): Biproximate Method with Analysis of Precision and Accuracy.

To review the anatomic basis of prostate boundary selection on T2-weighted magnetic resonance imaging (MRI). To introduce an alternative 3D ellipsoid measuring technique that maximizes precision, report the intra- and inter-observer reliability, and to advocate it's use for research involving multiple observers. We demonstrate prostate boundary anatomy using gross pathology and MRI examples. This provides background for selecting key boundary marks when measuring prostate volume. An alternative ellipsoid volume method is then proposed using these boundaries in an attempt to improve inter-observer precision. An IRB approved retrospective study of 140 patients with elevated serum prostate specific antigen levels and/or abnormal digital rectal examinations was done with T2-weighted MRI applying a new (Biproximate) technique. Measurements were made by 2 examiners, correlated with each other for inter-observer precision and correlated with an expert observer for accuracy. Correlation statistics, linear regression analysis, and tests of means were applied using p ≤ 0.05 as the threshold for significance. Inter-observer correlation (precision) was 0.95 between observers. Correlation between these observers and the expert (accuracy) was 0.94 and 0.97 respectively. Intra-observer correlation for the expert was 0.98. Means for inter-rater reliability and accuracy were all the same (p = 0.001). We conclude that using more precise reproducible landmarks with biproximate technique, precision and accuracy of total prostate volume is found to be demonstrated.

Precision and accuracy of magnetic resonance imaging for lobar classification of benign prostatic hyperplasia.

PURPOSE: To validate the application of a magnetic resonance imaging (MRI)-based lobar classification of benign prostatic hyperplasia (BPH) for use in research and clinical management. METHODS: Two radiologists with 5 and 11 years post-fellowship experience were trained in the lobar classification of BPH using an internally developed atlas of prostate anatomy with example MRI images edited by a third senior radiologist designated as the "administrator" of the study. A study population of 140 patients referred to a tertiary academic medical center with known or suspected prostate cancer was selected by the administrator to test the interrater reliability (IRR; precision) of the classification as well as accuracy of the two readers compared to the administrator as the "gold" standard. The intrarater reliability of repeat readings of the administrator was also examined. Percentage of agreement, proportion of agreement, and Cohen's κ were applied. This was a retrospective IRB-approved study. RESULTS: IRR (precision) between the two interpreting radiologists was 64% agreement, corresponding to unweighted κ of 0.52. Composite proportion of agreement across all BPH types (categories) for interpreting radiologists was 0.67. Observer accuracy was 62% agreement, unweighted κ 0.49, for observer 1 and 67%, unweighted κ 0.58, for observer 2. Intrarater reliability for the administrator was 87% agreement, unweighted κ 0.81 with composite proportion of agreement across all categories of 0.87. CONCLUSIONS: MRI lobar classification of BPH is a reproducible and reliable tool for research and clinical applications.

Use of MRI for Lobar Classification of Benign Prostatic Hyperplasia: Potential Phenotypic Biomarkers for Research on Treatment Strategies.

OBJECTIVE: We present an MRI classification of benign prostatic hyperplasia (BPH) for use as a phenotype biomarker in the study of proposed therapeutic interventions. CONCLUSION: Six patterns of BPH distribution were identified. Illustrations are provided for each classification type.

Imaging of non-neoplastic intratesticular masses.

The use of high-frequency ultrasound is increasing for the treatment of cystic, vascular, and solid non-neoplastic intratesticular masses. Cystic lesions examined include simple testicular cysts, tunica albuginea cysts, epidermoid cysts, tubular ectasia of rete testis, and intratesticular abscesses. Vascular lesions examined include intratesticular varicocele and intratesticular arteriovenous malformations. Solid lesions examined include fibrous pseudotumor of the testis, focal or segmental testicular infarct, fibrosis of the testis, testicular hematoma, congenital testicular adrenal rests, tuberculoma, and sarcoidosis. Gray-scale and color-flow Doppler sonography facilitate the visualization of the benign characteristics of the lesions. Magnetic resonance imaging can also help as a problem-solving modality in some cases.

Benign prostatic hyperplasia: a review and ultrasound classification.

Benign prostatic hyperplasia (BPH) is reviewed, including definition,natural history, diagnosis, and brief discussion of clinical management. Transrectal ultrasound is used to demonstrate the gross anatomy and natural history of BPH and its role in diagnosis and treatment strategies. A potentially useful ultrasound classification of BPH based on the gross anatomic classification of Randall is illustrated.

Elevation of dipeptidylpeptidase iv activities in the prostate peripheral zone and prostatic secretions of men with prostate cancer: possible prostate cancer disease marker.

PURPOSE: Dipeptidylpeptidase IV (DPIV) is a multifunctional type II plasma membrane glycoprotein with serine-type exopeptidase activity that is secreted by the prostate and increased in prostate cancer. We determined whether changes in DPIV activities in prostatic tissue zones and expressed secretions were associated with the presence of cancer. MATERIALS AND METHODS: Expressed prostatic secretion (EPS), and biopsy of the transition (TZ) and peripheral (PZ) zones were collected from men undergoing ultrasound guided prostate biopsy. DPIV activities were measured by glypro-p-nitroanalide hydrolysis. RESULTS: DPIV activities were significantly higher in TZ than in PZ tissues in men with no evidence of malignancy. However, activities in EPS were negatively associated with TZ volume and positively associated with PZ volume. Mean and median DPIV activities in EPS from men with biopsy determined cancer were significantly higher than in men with no evidence of malignancy. DPIV activities in TZ and PZ biopsies were higher in men with cancer but most markedly in the PZ. CONCLUSIONS: These data indicate that secreted DPIV originates from the TZ and PZ. Increased DPIV activities in cancer are strongly associated with the PZ, which is the zone most commonly involved with cancer. Measuring DPIV levels in expressed EPS or post-digital rectal prostate examination urine may be useful for evaluating men for prostate cancer.

Imaging of urethral disease: a pictorial review.

Retrograde urethrography and voiding cystourethrography are the modalities of choice for imaging the urethra. Cross-sectional imaging modalities, including ultrasonography, magnetic resonance (MR) imaging, and computed tomography, are useful for evaluating periurethral structures. Retrograde urethrography is the primary imaging modality for evaluating traumatic injuries and inflammatory and stricture diseases of the male urethra. Sonourethrography plays an important role in the assessment of the thickness and length of bulbar urethral stricture. Although voiding cystourethrography is frequently used to evaluate urethral diverticula in women, MR imaging is highly sensitive in the demonstration of these entities. MR imaging is also accurate in the local staging of urethral tumors.