AI in Oncology
Research and news where artificial intelligence is the method — diagnosis, drug discovery, pathology, clinical decision support.
AI-Assisted Three-Class Oral Histopathology: A Reader Study.
Cheng S, et al
A three-class deep learning model (VGG16-BR) trained on 25,576 histopathology patches achieved a test-set accuracy of 0.9065, macro-F1 of 0.9086, and macro-AUC of 0.9628 in distinguishing normal tissue, oral leukoplakia, and oral squamous cell carcinoma, with an OSCC-to-Normal misclassification rate…
Source →Measuring Radiologist Workload After AI Triage in Breast Cancer Screening.
Sorin V, et al
A new session-level reporting standard is proposed to more accurately quantify radiologist workload following AI triage in breast cancer screening, replacing simple reading counts with total active human interpretation minutes per 1,000 women screened. The authors argue that reading count alone fail…
Source →Interpretable four-class machine learning prediction of initial I-131 therapy responses in differentiated thyroid cancer.
Lei Y, et al
A four-class Random Forest machine learning framework achieved a micro-average AUC of 0.894 (95% CI: 0.842–0.901) in the test cohort and 0.885 (95% CI: 0.832–0.912) in temporal validation when predicting four distinct response categories — excellent response, indeterminate response, biochemical inco…
Source →A Transformer-Based Deep Learning Model for Prediction of Temozolomide Resistance in Glioblastoma Using Pretreatment MRI Images.
Xue C, et al
A Vision Transformer-based fusion model combining pretreatment MRI deep learning analysis with MGMT methylation status achieved AUC values of 0.855, 0.917, 0.806, and 0.848 across the training, internal validation, and two external validation cohorts, respectively, for predicting temozolomide (TMZ) …
Source →High-Content Imaging Workflow for Single-Cell Drug Response Profiling of Acute Myeloid Leukemia Subtypes.
Isigkeit L, et al
Researchers have developed a high-content fluorescence imaging platform that profiles drug responses in Acute Myeloid Leukaemia (AML) at the level of individual cells, overcoming the limitations of conventional bulk viability assays that mask tumour heterogeneity. The workflow integrates three non-t…
Source →Exploring the Molecular Mechanisms of Weifuchun in the Treatment of Chronic Atrophic Gastritis Based on Single-cell Sequencing and Machine Learning.
Chen H, et al
A study combining single-cell RNA sequencing and machine learning identified JUN and MMP12 as the key molecular targets through which Weifuchun (WFC), a classical traditional Chinese medicine compound, exerts its therapeutic effects against chronic atrophic gastritis (CAG) — a common precancerous le…
Source →Deep Learning for Automated Detection and Segmentation of Meningioma on Multiparametric MRI.
Farré-Melero A, et al
A deep learning model combining three MRI sequences — contrast-enhanced T1 (T1-CE), standard T1, and FLAIR — achieved a near-perfect meningioma detection rate of 0.98 ± 0.03 and a DICE segmentation score of 0.82 ± 0.04 in a retrospective study of 149 patients, substantially outperforming all single-…
Source →Topologically distinct intratumoral heterogeneity scores for predicting high-risk pathological grades in invasive lung adenocarcinoma: A multicenter study across four institutions.
Lin S, et al
A multicenter study of 1,051 patients with invasive lung adenocarcinoma (IAC) demonstrated that a stacking ensemble machine learning classifier combining topologically distinct intratumoral heterogeneity (ITH) scores derived from CT images with clinicoradiological features achieved an area under the…
Source →Interpretable prediction of macrotrabecular-massive HCC and recurrence-free survival by integrating MRI LI-RADS features with deep learning habitat radiomics.
Zhang M, et al
A multicenter study of 607 patients with early-stage hepatocellular carcinoma (HCC) demonstrated that a nomogram integrating MRI LI-RADS features, deep learning scores, and habitat radiomics can preoperatively identify the aggressive macrotrabecular-massive (MTM+) subtype with areas under the ROC cu…
Source →From cell counts to cellular interactions: Cu-Cyto and the co-localization index as a spatial framework for the tumor immune microenvironment of rectal cancer.
Yamashita K, et al
A new deep learning framework combining Cu-Cyto image cytometry — capable of detecting and classifying approximately twenty cell types from standard immunohistochemistry whole-slide images — with a novel Co-Localization Index demonstrates that stromal, but not intratumoral, density of CD103⁺CD8⁺ tis…
Source →This page is educational — it is not medical advice and does not replace consultation with an oncologist. Diagnostic and treatment decisions are made solely by specialist physicians.