Perceptual clustering for automatic hotspot detection from Ki‐67‐stained neuroendocrine tumour images

Abstract

Hotspot detection plays a crucial role in grading of neuroendocrine tumours of the digestive system. Hotspots are often detected manually from Ki-67-stained images, a practice which is tedious, irreproducible and error prone. We report a new method to segment Ki-67-positive nuclei from Ki-67-stained slides of neuroendocrine tumours. The method combines minimal graph cuts along with the multistate difference of Gaussians to detect the individual cells from images of Ki-67-stained slides. It, then, automatically defines the composite function, which is used to determine hotspots in neuroendocrine tumour slide images. We combine modified particle swarm optimization with message passing clustering to mimic the thought process of the pathologist during hotspot detection in neuroendocrine tumour slide images. The proposed method was tested on 55 images of size 10 × 5 K and resulted in an accuracy of 94.60%. The developed methodology can also be part of the workflow for other diseases such as breast cancer and glioblastomas.

An Advanced Deep Learning Approach for Ki-67 Stained Hotspot Detection and Proliferation Rate Scoring for Prognostic Evaluation of Breast Cancer

“…In comparison, the method in N. Khan et al . 31 requires an average of 7 seconds only to segment a color image. Overall, the proposed method is much more efficient than the existing Ki-67 scoring methods.…”

“…for which we cannot directly measure the performances of our proposed method with the existing methodologies. Based on some technical understanding of image similarities, we compared the qualitative and quantitative performances with the two recently published articles 15, 31 on Ki-67 scoring in Table 7. Furthermore, we measured the efficiency of our proposed framework with other conventional methods.…”

“…Khan et al . 31 Proposed MethodologyImage typeHuman nasopharyngeal carcinoma XenograftsNeuroendocrine tumorBreast cancerSample size100 images57 images450 imagesImage size2040 × 153610 × 5 K2048 × 1536Image Magnification40x40x40xMethodology usedConventional techniques (smoothing, color channel decomposition, local feature extraction, K-means, watershed segmentation)Conventional technique (Perceptual clustering) Deep Learning integrated with decision layer Accuracy (%)91.894.60 97 Computation time (sec)1.77 1.33 in GPU and 1.64 in CPU CPU or GPU usedCPUCPU CPU and GPU both Error rate0.82Not mentioned 0.41 …”

Tissue heterogeneity contributes to suboptimal precision of WHO 2010 scoring criteria for Ki67 labeling index in a subset of neuroendocrine neoplasms of the pancreascriteria for Ki67 labeling index in a subset of neuroendocrine neoplasms of the pancreas

“…Important aspect of Ki67 scoring is selection of tumour regions for counting [42,53,73,74,75,76,77,78]. Current counting methodology in NEN is based on recognition of HS.…”

“…HS selection is done subjectively and results of these choices (at least in breast cancer) do not necessarily correlated with HS recognized using digital image analysis [73,74]. Automated HS detection may be an alternative [74,76,77,79,80]. It is not fully clear whether single HS is enough for Ki67 LI assessment or whether several HS should be counted [81].…”