Computational Grading of GI-NETs » Computational Microscopy Imaging Laboratory » College of Medicine

We developed an automated Ki-67 index estimation method for gastrointestinal neuroendocrine tumors (GI-NETs) for defining the tumor grade.Toward this goal, we introduced two computational tools that utilize Ki-67 and synaptophysin double-immunostained (DS) slides to improve the accuracy of Ki-67 index quantitation in GI-NETs: (1) Synaptophysin-KI-Estimator (SKIE), a pipeline automating Ki-67 index quantitation via whole-slide image (WSI) analysis and (2) deep-SKIE, a deep learner-based approach where a Ki-67 index heatmap is generated throughout the tumor. The proposed methods can improve accuracy and potentially save a significant amount of time if implemented into clinical practice.

**Schematic diagram for Synaptophyin-Ki-67 Index Estimator (SKIE).** (a) Whole-slide image (WSI) of gastrointestinal neuroendocrine tumor tissue section stained with hematoxylin and eosin (H&E). (b) WSI of the adjacent tissue section stained with synaptophysin (red) and Ki-67 (brown) (or DS WSI). (c) Result of image registration by matching manually selected landmarks within 1a and 1b. (d, e) Binary mask of synaptophysin positive region and Ki-67 positive cells, respectively, obtained upon color deconvolution and morphological processing. (f) Automated detection of five candidate hot-spots containing the highest density of Ki-67 positive cells within tumor regions. (g) Selected candidate hot-spots chosen from the registered images. (h, i) Extracted hot-spots from the H&E and the double-immunostained images, respectively. (j) Overlay of nuclei mask obtained via unsupervised classification of pixels via k-means clustering of (h) to obtain all cells within tumor regions. (k) Overlay of the hot-spot using masks from (d) and (e) to obtain Ki-67 positive cells (highlighted in green) within tumor regions (bounded in blue).

Select References

D. Govind, K. Y. Jen, K. Matsukuma, G. Gao, K. A. Olson, D. Gui, G. Wilding, S. P. Border, and P. Sarder, “Improving the accuracy of gastrointestinal neuroendocrine tumor grading with deep learning,” Scientific Reports – Nature, vol. 10, pp. 11064: 1–11, Jul. 2020.

Underline indicates corresponding author.
^* indicates equal contribution.
^Ω indicates Dr. Sarder’s faculty trainees.
^§ indicates post doctoral associates.
^† indicates Dr. Sarder’s graduate students.
^‡ indicates Dr. Sarder’s undergraduate students.