Computer-driven quantitative image analysis in the assessment of tumor cell and T cell features in diffuse large B cell lymphomas.
Francesco GaudioRoberto TammaGiuseppe IngravalloTommasina PerroneFilomena Emanuela LaddagaMariastella De CandiaEugenio MaioranoDomenico RibattiGiorgina SpecchiaPublished in: Annals of hematology (2018)
Diffuse large B cell lymphoma (DLBCL) is recognized as the most common non-Hodgkin lymphoma subtype. Advanced high-resolution digital scans of pathology slides have enabled the development of computer-based image analysis algorithms that may assist pathologists in quantifying immunohistochemical stains. In this retrospective study, we reviewed data from 29 patients affected by DLBCL. In order to evaluate the number of tumor cells and microenvironment T cells, we performed an analysis of CD20, Ki67, and CD3 counts, assessed with the Positive Pixel Count algorithm embedded in the Aperio ImageScope software. A lower tumor cell count was observed in patients with a non-germinal center immunophenotype, high LDH, splenomegaly and an IPI ≥ 3. A lower number of CD3 was observed in patients with bulky disease, an IPI ≥ 3 and disease stage 3-4. Overall, these data confirm that quantitative analysis of the tumor cells and of the tumor microenvironment by means of computer-driven quantitative image analysis may add new information in DLBCL diagnosis.
Keyphrases
- diffuse large b cell lymphoma
- high resolution
- deep learning
- epstein barr virus
- machine learning
- single cell
- end stage renal disease
- chronic kidney disease
- cell therapy
- electronic health record
- big data
- peripheral blood
- ejection fraction
- stem cells
- newly diagnosed
- computed tomography
- nk cells
- artificial intelligence
- data analysis
- mass spectrometry
- peritoneal dialysis
- low grade
- neoadjuvant chemotherapy
- healthcare
- magnetic resonance
- bone marrow
- tandem mass spectrometry
- high speed
- contrast enhanced