Automatic Prostate Gleason Grading Using Pyramid Semantic Parsing Network in Digital Histopathology

This article provides a status report on the global burden of cancer worldwide using the GLOBOCAN 2018 estimates of cancer incidence and mortality produced by the International Agency for Research on Cancer, with a focus on geographic variability across 20 world regions. There will be an estimated 18.1 million new cancer cases (17.0 million excluding nonmelanoma skin cancer) and 9.6 million cancer deaths (9.5 million excluding nonmelanoma skin cancer) in 2018. In both sexes combined, lung cancer is the most commonly diagnosed cancer (11.6% of the total cases) and the leading cause of cancer death (18.4% of the total cancer deaths), closely followed by female breast cancer (11.6%), prostate cancer (7.1%), and colorectal cancer (6.1%) for incidence and colorectal cancer (9.2%), stomach cancer (8.2%), and liver cancer (8.2%) for mortality. Lung cancer is the most frequent cancer and the leading cause of cancer death among males, followed by prostate and colorectal cancer (for incidence) and liver and stomach cancer (for mortality). Among females, breast cancer is the most commonly diagnosed cancer and the leading cause of cancer death, followed by colorectal and lung cancer (for incidence), and vice versa (for mortality); cervical cancer ranks fourth for both incidence and mortality. The most frequently diagnosed cancer and the leading cause of cancer death, however, substantially vary across countries and within each country depending on the degree of economic development and associated social and life style factors. It is noteworthy that high-quality cancer registry data, the basis for planning and implementing evidence-based cancer control programs, are not available in most low- and middle-income countries. The Global Initiative for Cancer Registry Development is an international partnership that supports better estimation, as well as the collection and use of local data, to prioritize and evaluate national cancer control efforts. CA: A Cancer Journal for Clinicians 2018;0:1-31. © 2018 American Cancer Society.

We present a novel and practical deep fully convolutional neural network architecture for semantic pixel-wise segmentation termed SegNet. This core trainable segmentation engine consists of an encoder network, a corresponding decoder network followed by a pixel-wise classification layer. The architecture of the encoder network is topologically identical to the 13 convolutional layers in the VGG16 network [1] . The role of the decoder network is to map the low resolution encoder feature maps to full input resolution feature maps for pixel-wise classification. The novelty of SegNet lies is in the manner in which the decoder upsamples its lower resolution input feature map(s). Specifically, the decoder uses pooling indices computed in the max-pooling step of the corresponding encoder to perform non-linear upsampling. This eliminates the need for learning to upsample. The upsampled maps are sparse and are then convolved with trainable filters to produce dense feature maps. We compare our proposed architecture with the widely adopted FCN [2] and also with the well known DeepLab-LargeFOV [3] , DeconvNet [4] architectures. This comparison reveals the memory versus accuracy trade-off involved in achieving good segmentation performance. SegNet was primarily motivated by scene understanding applications. Hence, it is designed to be efficient both in terms of memory and computational time during inference. It is also significantly smaller in the number of trainable parameters than other competing architectures and can be trained end-to-end using stochastic gradient descent. We also performed a controlled benchmark of SegNet and other architectures on both road scenes and SUN RGB-D indoor scene segmentation tasks. These quantitative assessments show that SegNet provides good performance with competitive inference time and most efficient inference memory-wise as compared to other architectures. We also provide a Caffe implementation of SegNet and a web demo at http://mi.eng.cam.ac.uk/projects/segnet.

In November, 2014, 65 prostate cancer pathology experts, along with 17 clinicians including urologists, radiation oncologists, and medical oncologists from 19 different countries gathered in a consensus conference to update the grading of prostate cancer, last revised in 2005. The major conclusions were: (1) Cribriform glands should be assigned a Gleason pattern 4, regardless of morphology; (2) Glomeruloid glands should be assigned a Gleason pattern 4, regardless of morphology; (3) Grading of mucinous carcinoma of the prostate should be based on its underlying growth pattern rather than grading them all as pattern 4; and (4) Intraductal carcinoma of the prostate without invasive carcinoma should not be assigned a Gleason grade and a comment as to its invariable association with aggressive prostate cancer should be made. Regarding morphologies of Gleason patterns, there was clear consensus on: (1) Gleason pattern 4 includes cribriform, fused, and poorly formed glands; (2) The term hypernephromatoid cancer should not be used; (3) For a diagnosis of Gleason pattern 4, it needs to be seen at 10x lens magnification; (4) Occasional/seemingly poorly formed or fused glands between well-formed glands is insufficient for a diagnosis of pattern 4; (5) In cases with borderline morphology between Gleason pattern 3 and pattern 4 and crush artifacts, the lower grade should be favored; (6) Branched glands are allowed in Gleason pattern 3; (7) Small solid cylinders represent Gleason pattern 5; (8) Solid medium to large nests with rosette-like spaces should be considered to represent Gleason pattern 5; and (9) Presence of unequivocal comedonecrosis, even if focal is indicative of Gleason pattern 5. It was recognized by both pathologists and clinicians that despite the above changes, there were deficiencies with the Gleason system. The Gleason grading system ranges from 2 to 10, yet 6 is the lowest score currently assigned. When patients are told that they have a Gleason score 6 out of 10, it implies that their prognosis is intermediate and contributes to their fear of having a more aggressive cancer. Also, in the literature and for therapeutic purposes, various scores have been incorrectly grouped together with the assumption that they have a similar prognosis. For example, many classification systems consider Gleason score 7 as a single score without distinguishing 3+4 versus 4+3, despite studies showing significantly worse prognosis for the latter. The basis for a new grading system was proposed in 2013 by one of the authors (J.I.E.) based on data from Johns Hopkins Hospital resulting in 5 prognostically distinct Grade Groups. This new system was validated in a multi-institutional study of over 20,000 radical prostatectomy specimens, over 16,000 needle biopsy specimens, and over 5,000 biopsies followed by radiation therapy. There was broad (90%) consensus for the adoption of this new prostate cancer Grading system in the 2014 consensus conference based on: (1) the new classification provided more accurate stratification of tumors than the current system; (2) the classification simplified the number of grading categories from Gleason scores 2 to 10, with even more permutations based on different pattern combinations, to Grade Groups 1 to 5; (3) the lowest grade is 1 not 6 as in Gleason, with the potential to reduce overtreatment of indolent cancer; and (4) the current modified Gleason grading, which forms the basis for the new grade groups, bears little resemblance to the original Gleason system. The new grades would, for the foreseeable future, be used in conjunction with the Gleason system [ie. Gleason score 3+3=6 (Grade Group 1)]. The new grading system and the terminology Grade Groups 1-5 have also been accepted by the World Health Organization for the 2016 edition of Pathology and Genetics: Tumours of the Urinary System and Male Genital Organs.