Around the world in a million cells
By using present single-cell morphological and topological profiling to characterize cells on whole-slide images (WSIs), a team of researchers have created a comprehensive atlas of human breast cancer (1). With over 410 million cells included, the atlas reveals “the phenotypic diversity of [the] breast cancer ecosystem on multiple levels,” according to the paper. It’s hoped that the multiomics data gleaned from the atlas will have clinical applications, potentially defining future disease biomarkers.
Glass is the past
A comparison between WSIs and traditional glass slides has shown positive diagnostic applications for digital analysis in cytopathology for cervical cancer. Using 99 cases of vaginal cytology, a team of researchers have established an agreement between digital and optical microscopy in a number of different categories – particularly in the cases of inflammatory processes and cellular atypia (2).
Scanning for survival
A new study in Lancet Oncology has explored the feasibility of AI application in cases of hepatocellular carcinoma that present with atezolizumab–bevacizumab response signature (ABRS) – which is associated with positive patient outcomes (3). The researchers found that AI trained on a dataset built from The Cancer Genome Atlas could estimate ABRS directly from histological slides and could accurately determine progression-free survival.
Prying into prion
Researchers from Italy have established an effective AI-powered system for the detection of prion disease in WSIs. The first study to implement such methods in prion disease, the research also saw the creation of a novel image format – DeltaE transform – capable of boosting machine learning performance and improving analysis of morphological features (4).
Viral virulence
Analyses of the virulence of SARS-CoV-2 variants in hamster models show enhancement by digital pathology and machine learning methods, helping create an online repository of whole-organ histopathology (5).
Deep learning in cancer
Deep learning models create a searchable digital atlas of 35 breast cancer tumor types. Such atlases could be used as a computational second opinion in diagnoses (6).
Colorectal shotgun
Digital spatial profiling partnered with shotgun proteomics to investigate the mechanisms of colorectal cancer metastasis (7).
Collaborative imaging
New cloud-based digital pathology platform incorporates an image annotation and analysis framework, developed via an iterative active learning process (8).
Newsletters
Receive the latest pathology news, personalities, education, and career development – weekly to your inbox.
References
- S Zhao et al., “Single-cell morphological and topological atlas reveals the ecosystem diversity of human breast cancer,” Nat Commun, 14, 6796 (2023). PMID: 37880211.
- G de Velozo et al., “Comparison of glass and digital slides for cervical cytopathology screening and interpretation,” Diagn Cytopathol, 51, 735 (2023). PMID: 37587842.
- Q Zeng et al., “Artificial intelligence-based pathology as a biomarker of sensitivity to atezolizumab–bevacizumab in patients with hepatocellular carcinoma: a multicentre retrospective study,” Lancet Oncol (2023). PMID: 37951222.
- M Salvi et al., “Quantitative analysis of prion disease using an AI-powered digital pathology framework,” Sci Rep, 13, 17759 (2023). PMID: 37853094.
- GR Meehan et al., “Phenotyping the virulence of SARS-CoV-2 variants in hamsters by digital pathology and machine learning,” PLOS Pathogens, 19, e1011589 (2023). PMID: 37934791.
- A Shafique et al., “A Preliminary Investigation into Search and Matching for Tumour Discrimination in WHO Breast Taxonomy Using Deep Networks,” Mod Pathol (2023). PMID: 37939901.
- NPM Brouwer et al., “Transcriptomics and proteomics reveal distinct biology for lymphnode metastases and tumour deposits in colorectal cancer,” J Pathol, 261, 401 (2023). PMID: 37792663.
- R Jesus et al., “Personalizable AI platform for universal access to research and diagnosis in digital pathology,” Comput Methods Programs Biomed, 242, 107787 (2023). PMID: 37717524.
