https://www.ncbi.nlm.nih.gov/pubmed/29795581
High-throughput adaptive sampling for whole-slide histopathology image analysis (HASHI) via convolutional neural networks: Application to invasive breast cancer detection.
Author information
- 1
- School of Engineering, Universidad de los Llanos, Villavicencio, Meta, Colombia.
- 2
- Dept. of Computing Systems and Industrial Engineering, Universidad Nacional de Colombia, Bogotá, Cundinamarca, Colombia.
- 3
- University Hospitals Case Medical Center, Cleveland, OH, United States of America.
- 4
- Inspirata Inc., Tampa, FL, United States of America.
- 5
- Hospital of the University of Pennsylvania, Philadelphia, PA, United States of America.
- 6
- Cancer Institute of New Jersey, New Brunswick, NJ, United States of America.
- 7
- University at Buffalo, The State University of New York, Buffalo, NY, United States of America.
- 8
- Case Western Reserve University, Cleveland, OH, United States of America.
Abstract
Precise detection of invasive cancer on whole-slide images (WSI) is a critical first step in digital pathology tasks of diagnosis and grading. Convolutional neural network (CNN) is the most popular representation learning method for computer vision tasks, which have been successfully applied in digital pathology, including tumor and mitosis detection. However, CNNs are typically only tenable with relatively small image sizes (200 × 200 pixels). Only recently, Fully convolutional networks (FCN) are able to deal with larger image sizes (500 × 500 pixels) for semantic segmentation. Hence, the direct application of CNNs to WSI is not computationally feasible because for a WSI, a CNN would require billions or trillions of parameters. To alleviate this issue, this paper presents a novel method, High-throughput Adaptive Sampling for whole-slide Histopathology Image analysis (HASHI), which involves: i) a new efficient adaptive sampling method based on probability gradient and quasi-Monte Carlo sampling, and, ii) a powerful representation learning classifier based on CNNs. We applied HASHI to automated detection of invasive breast cancer on WSI. HASHI was trained and validated using three different data cohorts involving near 500 cases and then independently tested on 195 studies from The Cancer Genome Atlas. The results show that (1) the adaptive sampling method is an effective strategy to deal with WSI without compromising prediction accuracy by obtaining comparative results of a dense sampling (∼6 million of samples in 24 hours) with far fewer samples (∼2,000 samples in 1 minute), and (2) on an independent test dataset, HASHI is effective and robust to data from multiple sites, scanners, and platforms, achieving an average Dice coefficient of 76%.
- PMID:
- 29795581
- PMCID:
- PMC5967747
- DOI:
- 10.1371/journal.pone.0196828
유방암 병리 슬라이드 전체(이미지: 50k *50k pixels) 를 보는(HASHI) FCN.
4개 연구기관 자료 종합.
"adaptive sampling method "
"quasi-Monte-Carlo sampling with gradient-based adaptive strategy" : uncertainty 가 높은 영역을 집중적으로 보는 방법
24시간동안, 약 600만개의 샘플
GPU : Titan X with 12GB memory
한계점 : 초기암 소견인 DCIS 를 구분해 내지는 못했다.
유방암 구체적 병리 소견인, cacner grading, tubule and mitosis counting 등의 추가적인 섬세한 일을 하지는 못한다.
의의 : 전체 병리 슬라이드를 보는, 병리의사 (병리과)의 일을 보조하는데 (to speed up) 는 도움이 될 듯. / 대신 computation hardware 셋팅비용, software 유지보수, false positive 등을 관리하는 데 시간과 비용과 노력은 더 들지 않을까?
pdf : http://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0196828&type=printable
( > 10.0 Mb )
