Data Augmentation and Transfer Learning to Improve Generalizability of an Automated Prostate Segmentation Model

Sanford, Thomas; Zhang, Ling; Harmon, Stephanie; Sackett, Jonathan; Yang, Dong; Roth, Holger R.; Xu, Ziyue; Kesani, Deepak; Mehralivand, Sherif; Baroni, Ronaldo Hueb; Barrett, Tristan; Girometti, Rossano; Oto, Aytekin; Purysko, Andrei S.; Xu, Sheng; Pinto, Peter A.; Xu, Daguang; Wood, Bradford J.; Choyke, Peter L.; Türkbey, Barış

doi:10.2214/ajr.19.22347

Cited by 40 publications

(34 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Lee et al reported a DSC of 87% for whole-gland segmentation and 77% for the TZ, while Sanford obtained 93% and 89%, respectively. 17,18 However, these authors employed computationally expensive DL networks trained on cloud-based hardware, also requiring highspeed Internet availability. Findings of the present study show that smaller, easier to train and less computationally expensive networks like ENet can achieve similar results to state-of-theart solutions.…”

Section: Discussionmentioning

confidence: 99%

“…Larger datasets have been used for prostate segmentation in other studies; however, the choice to employ a public dataset allows easier reproducibility of findings as well as a clear benchmark for future studies in this research space. 17,18 In the same vein, all masks are freely available to use by other researchers. To ensure a high quality of the ground truth annotations, a consensus approach by multiple readers was employed.…”

Section: Limitationsmentioning

confidence: 99%

“…[14][15][16] Among the tasks that can be addressed by ML and DL is that of automated prostate segmentation, as shown in previous works. 17,18 However, DL networks often require high computational power and training time, limiting clinical applicability. 19,20 The aim of this study was to compare different DL networks in the task of automated whole-gland and zonal anatomy segmentation of prostate MRI exams, taking into account their accuracy, training time, and hardware requirements.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Deep Learning Whole‐Gland and Zonal Prostate Segmentation on a Public MRI Dataset

Cuocolo

Comelli

Stefano

et al. 2021

Magnetic Resonance Imaging

View full text Add to dashboard Cite

Background: Prostate volume, as determined by magnetic resonance imaging (MRI), is a useful biomarker both for distinguishing between benign and malignant pathology and can be used either alone or combined with other parameters such as prostate-specific antigen.Purpose: This study compared different deep learning methods for whole-gland and zonal prostate segmentation. Study Type: Retrospective. Population: A total of 204 patients (train/test = 99/105) from the PROSTATEx public dataset. Field strength/Sequence: A 3 T, TSE T 2 -weighted. Assessment: Four operators performed manual segmentation of the whole-gland, central zone + anterior stroma + transition zone (TZ), and peripheral zone (PZ). U-net, efficient neural network (ENet), and efficient residual factorized ConvNet (ERFNet) were trained and tuned on the training data through 5-fold cross-validation to segment the whole gland and TZ separately, while PZ automated masks were obtained by the subtraction of the first two. Statistical Tests: Networks were evaluated on the test set using various accuracy metrics, including the Dice similarity coefficient (DSC). Model DSC was compared in both the training and test sets using the analysis of variance test (ANOVA) and post hoc tests. Parameter number, disk size, training, and inference times determined network computational complexity and were also used to assess the model performance differences. A P < 0.05 was selected to indicate the statistical significance. Results: The best DSC (P < 0.05) in the test set was achieved by ENet: 91% AE 4% for the whole gland, 87% AE 5% for the TZ, and 71% AE 8% for the PZ. U-net and ERFNet obtained, respectively, 88% AE 6% and 87% AE 6% for the whole gland, 86% AE 7% and 84% AE 7% for the TZ, and 70% AE 8% and 65 AE 8% for the PZ. Training and inference time were lowest for ENet. Data Conclusion: Deep learning networks can accurately segment the prostate using T 2 -weighted images. Evidence Level: 4 Technical Efficacy: Stage 2

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Limitationsmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Deep Learning Whole‐Gland and Zonal Prostate Segmentation on a Public MRI Dataset

Cuocolo

Comelli

Stefano

et al. 2021

Magnetic Resonance Imaging

View full text Add to dashboard Cite

show abstract

“…Third, ProGNet performed better than or equal to other prostate segmentation models. [10][11][12][13][14][16][17][18][19][20] The generalizability of ProGNet results from the large training (n=805) and testing (n=167) cohorts. Prior publications typically included only 40-250 cases.…”

Section: Discussionmentioning

confidence: 99%

“…7 Achieving generalizable results requires large amounts of training data from multiple institutions. 8,9 Different methods have been proposed to automate prostate gland segmentation [10][11][12][13][14][15][16][17][18][19][20][21] but have often used small datasets (usually 40-250 cases) [10][11][12][13][14][15][16][17][18] , did not use volumetric context from adjacent T2-MRI slices to make predictions 15,16 , failed to evaluate on external cohorts 11,18,19 , solely used single-institution training sets 11,[18][19][20] , did not release code for comparison [10][11][12]14,15,[17][18][19][20][21] or did not publish model accuracy. 21 Deep learning for medical applications has rarely-and never for the essential prostate segmentation task-been integrated into clinical practice, while reporting results and releasing the code online.…”

Section: Introductionmentioning

confidence: 99%

Deep Learning Improves Speed and Accuracy of Prostate Gland Segmentations on Magnetic Resonance Imaging for Targeted Biopsy

et al. 2021

View full text Add to dashboard Cite

Targeted biopsy improves prostate cancer diagnosis. Accurate prostate segmentation on MRI is critical for accurate biopsy. Manual gland segmentation is tedious and time-consuming. We sought to develop a deep learning model to rapidly and accurately segment the prostate on MRI and to implement it as part of routine MR-US fusion biopsy in the clinic.Materials and methods: 905 subjects underwent multiparametric MRI at 29 institutions, followed by MR-US fusion biopsy at one institution. A urologic oncology expert segmented the prostate on axial T2-weighted MRI scans. We trained a deep learning model, ProGNet, on 805 cases. We retrospectively tested ProGNet on 100 independent internal and 56 external cases. We prospectively implemented ProGNet as part of the fusion biopsy procedure for 11 patients. We compared ProGNet performance to two deep learning networks (U-Net and HED) and radiology technicians. The Dice similarity coefficient (DSC) was used to measure overlap with expert segmentations. DSCs were compared using paired t-tests. Results: ProGNet (DSC=0.92) outperformed U-Net (DSC=0.85, p<0.0001), HED (DSC=0.80, p<0.0001), and radiology technicians (DSC=0.89, p<0.0001) in the retrospective internal test set. In the prospective cohort, ProGNet (DSC=0.93) outperformed radiology technicians (DSC=0.90, p<0.0001). ProGNet took just 35 seconds per case (vs. 10 minutes for radiology technicians) to yield a clinically utilizable segmentation file. Conclusions: This is the first study to employ a deep learning model for prostate gland segmentation for targeted biopsy in routine urologic clinical practice, while reporting results

show abstract

Multi-institutional Investigation of Model Generalizability for Virtual Contrast-Enhanced MRI Synthesis

Lam

Liu

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Data Augmentation and Transfer Learning to Improve Generalizability of an Automated Prostate Segmentation Model

Cited by 40 publications

References 11 publications

Deep Learning Whole‐Gland and Zonal Prostate Segmentation on a Public MRI Dataset

Deep Learning Whole‐Gland and Zonal Prostate Segmentation on a Public MRI Dataset

Deep Learning Improves Speed and Accuracy of Prostate Gland Segmentations on Magnetic Resonance Imaging for Targeted Biopsy

Multi-institutional Investigation of Model Generalizability for Virtual Contrast-Enhanced MRI Synthesis

Contact Info

Product

Resources

About