Automated T1 bladder risk stratification based on depth of lamina propria invasion from H and E tissue biopsies: a deep learning approach

Identification of bladder layers from tissue biopsies is the first step towards an accurate diagnosis and prognosis of bladder cancer. We present an automated Bladder Image Analysis System (BLIAS) that can recognize urothelium, lamina propria, and muscularis propria from images of H and E-stained slides of bladder biopsies. Furthermore, we present its clinical application to automate risk stratification of T1 bladder cancer patients based on the depth of lamina propria invasion. The method uses multidimensional scaling and transfer learning in conjunction with convolutional neural networks to identify different bladder layers from H and E images of bladder biopsies. The method was trained and tested on eighty whole slide images of bladder cancer biopsies. Our preliminary findings suggest that the proposed method has good agreement with the pathologist in identification of different bladder layers. Additionally, given a set of tumor nuclei within lamina propria, it has the potential to risk stratify T1 bladder cancer by computing the distance from this set to urothelium and muscularis propria. Our results suggest that a pretrained network trained via transfer learning is better in identifying bladder layers than a conventional deep learning paradigm.

[1]  Steven L. Chang,et al.  Improving selection criteria for early cystectomy in high-grade t1 bladder cancer: a meta-analysis of 15,215 patients. , 2015, Journal of clinical oncology : official journal of the American Society of Clinical Oncology.

[2]  J. Epstein,et al.  An objective morphologic parameter to aid in the diagnosis of flat urothelial carcinoma in situ. , 2001, Human pathology.

[3]  A. Ruifrok,et al.  Quantification of histochemical staining by color deconvolution. , 2001, Analytical and quantitative cytology and histology.

[4]  Liang Cheng,et al.  Staging and reporting of urothelial carcinoma of the urinary bladder , 2009, Modern Pathology.

[5]  M. Khalid Khan,et al.  Entropy based quantification of Ki-67 positive cell images and its evaluation by a reader study , 2013, Medical Imaging.

[6]  M. Khalid Khan,et al.  An application of transfer learning to neutrophil cluster detection for tuberculosis: efficient implementation with nonmetric multidimensional scaling and sampling , 2018, Medical Imaging.

[7]  C. Compton,et al.  AJCC Cancer Staging Manual , 2002, Springer New York.

[8]  B. Inman,et al.  Perioperative intravesical chemotherapy in non-muscle-invasive bladder cancer: a systematic review and meta-analysis. , 2013, Journal of the National Comprehensive Cancer Network : JNCCN.

[9]  J. Palou,et al.  Multivariate analysis of the prognostic factors of primary superficial bladder cancer. , 2000, The Journal of urology.

[10]  W. Murphy,et al.  Accurate pathological staging of urothelial neoplasms requires better cystoscopic sampling. , 2002, The Journal of urology.

[11]  V. Boddi,et al.  Can early single dose instillation of epirubicin improve bacillus Calmette-Guerin efficacy in patients with nonmuscle invasive high risk bladder cancer? Results from a prospective, randomized, double-blind controlled study. , 2008, The Journal of urology.

[12]  M. Babjuk,et al.  Transurethral Resection of Non–muscle-invasive Bladder Cancer , 2009 .

[13]  J. Witjes,et al.  NMIBC risk calculators: how useful are they for the practicing urologist and how can their clinical utility be improved? , 2013, The Urologic clinics of North America.

[14]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[15]  Boqian Wu,et al.  FF-CNN: An Efficient Deep Neural Network for Mitosis Detection in Breast Cancer Histological Images , 2017, MIUA.

[16]  L. True,et al.  The usefulness of the level of the muscularis mucosae in the staging of invasive transitional cell carcinoma of the urinary bladder , 1990, Cancer.

[17]  Sam S. Chang,et al.  Challenges in the pathology of non-muscle-invasive bladder cancer: a dialogue between the urologic surgeon and the pathologist. , 2013, Urology.

[18]  M. Khalid Khan,et al.  A computational framework to detect normal and tuberculosis infected lung from H and E-stained whole slide images , 2017, Medical Imaging.

[19]  Pascal Vincent,et al.  Stacked Denoising Autoencoders: Learning Useful Representations in a Deep Network with a Local Denoising Criterion , 2010, J. Mach. Learn. Res..

[20]  P D Abel,et al.  Differing interpretations by pathologists of the pT category and grade of transitional cell cancer of the bladder. , 1988, British journal of urology.

[21]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[22]  J. Y. Lee,et al.  Prognostic Significance of Substaging according to the Depth of Lamina Propria Invasion in Primary T1 Transitional Cell Carcinoma of the Bladder , 2012, Korean journal of urology.

[23]  G. Sauter,et al.  Clinical significance of interobserver differences in the staging and grading of superficial bladder cancer , 2000, BJU international.

[24]  M. Babjuk,et al.  EAU guidelines on non-muscle-invasive urothelial carcinoma of the bladder: update 2013. , 2013, European urology.

[25]  Sergey Ioffe,et al.  Rethinking the Inception Architecture for Computer Vision , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).