ORCID

0009-0006-1789-7098

Date of Award

2025

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Pathology

First Advisor

Douglas J. Taatjes

Abstract

Computer-aided detection or diagnosis systems, which are deployed to detect abnormalities in histological samples, can play a key role in detecting a range of anomalies found in histologic samples. In fact, the earliest venture into the use of computerized image analysis and digital image processing was to analyze microscopy images, not in face recognition/detection, as many would claim. This automated image analysis continues to be transformative for many areas, including in areas of research and medicine like pathology and promises to offer highly accurate and efficient qualitative and quantitative analysis algorithms.

Due to these potentials, these algorithms and other related automated analysis techniques have attracted the attention of research in the field. These are deployed in a wide range of research topics, including cancer detection, classification, and even monitoring management response. This study aimed to provide a technical assessment of two analysis approaches that are utilized in software platforms: positive pixel count per tissue area analysis and cell-by-cell analysis approach. These two are commonly used in many digital histopathological slide analysis software(s) in research settings, including ImageScope and HALO, that respectively deploy the above algorithms and thus were used as proxies for this study.

We found that both the algorithms we worked with could only characterize IHC markers to some extent, HALO more so than ImageScope. The complicated nature of IHC assessment criteria, guided by the biology of the markers was the biggest influence on the challenges in image analysis. These need to be considered during algorithm development to obtain reliable automated analysis systems applicable in real-world research and/or hospital environments.

Language

en

Number of Pages

63 p.

Available for download on Tuesday, December 02, 2025

Included in

Pathology Commons

Share

COinS