AI in Oncology
Research and news where artificial intelligence is the method — diagnosis, drug discovery, pathology, clinical decision support.
Single-cell and spatial transcriptomic profiling identifies a fibroblast-like endothelial state and the MDK-NCL signaling axis in prostate cancer microenvironment remodeling.
Wang R, et al
Researchers have identified a fibroblast-like endothelial (FLE) cell population within the prostate cancer tumor microenvironment and established the MDK-NCL signaling axis as a key driver of intercellular communication and adverse clinical outcomes. Using integrated single-cell RNA sequencing, spat…
Source →Multi-omics, machine learning, and molecular simulation identify CD44 as a candidate target in endocrine-disrupting chemical-associated thyroid cancer progression.
Hu Y, et al
A multi-omics and machine learning study has identified CD44 as a key molecular target linking environmental endocrine-disrupting chemicals (EDCs) — including BPA, PFOA, and DEHP — to thyroid cancer progression, with a six-gene diagnostic model (FN1, BCL2, CD44, CDKN1A, CTNNB1, and JUN) achieving an…
Source →Automated artificial intelligence performance for longitudinal pulmonary nodule matching in lung cancer screening
In 361 participants of the UK Lung Cancer Screening trial undergoing 3-month follow-up low-dose CT, an AI nodule analysis system correctly matched 83.5% of persisting nodules to their baseline counterpart (283/339; 95% CI 79.2-87.1%). Accuracy depended strongly on how many nodules a participant had:…
Source →Emerging frontiers and challenges of artificial intelligence in PSMA-PET imaging: pioneering a new chapter in prostate cancer care.
Wang Y, et al
A comprehensive review published in the Japanese Journal of Radiology concludes that artificial intelligence — specifically radiomics and deep learning — holds transformative potential for PSMA-PET imaging in prostate cancer management, addressing persistent limitations such as interpretive subjecti…
Source →Utility of artificial intelligence for identifying thoracic lymphadenopathy in lung cancer
Muhiddin Dervis
In a multicentre study of 169 patients with operable thoracic malignancies, a commercial AI system classified 3,520 candidate thoracic lymph nodes on CT by nodal station with 96.1% sensitivity and 99.7% specificity, adjudicated against expert thoracic radiologists. Among the 169 nodes that were subs…
Source →Brain Tumor Classification from MRI Using Colormap-Based Vision Transformers on the BRISC2025 Dataset.
Ahmed F
A new AI framework called ViT-Color achieves 98.63% accuracy, 98.65% precision, 98.81% recall, an F1-score of 98.65%, and an AUC of 99.95% in classifying brain tumors from MRI scans by transforming grayscale images into pseudo-color representations before analysis with a Vision Transformer model. Th…
Source →A Multiparametric Ultrasound Diagnostic Model for Adenomyosis by Integrating Cervical Ultra-Microangiography and Shear Wave Elastography: A Prospective Study.
Xiao L, et al
A noninvasive multiparametric ultrasound model combining cervical shear wave elastography and ultra-microangiography achieved an area under the curve (AUC) of 0.94 in the training cohort and 0.96 in the validation cohort for diagnosing adenomyosis, outperforming individual parameters such as CIO_AL_…
Source →Capturing fine-grained spatial details in brain tumor MRI via multi-module feature enhancement.
Chen J, et al
A novel deep learning framework called FDANet achieved a mean average precision (mAP50) of 0.961 in MRI-based detection of brain tumors — including pituitary adenomas, gliomas, and meningiomas — representing a 6.0 percentage point improvement over the baseline YOLOv11 model while maintaining low par…
Source →MRI-based radiomics for non-invasive prediction of histopathological and immunohistochemical features in ductal carcinoma in situ: Implications for risk stratification.
Tan AA, et al
A retrospective study of 75 DCIS patients (74 women; mean age 49.36 ± 11.28 years) demonstrated that MRI-based radiomics can non-invasively predict key immunohistochemical and histopathological features of ductal carcinoma in situ with high accuracy, achieving F1 scores of 0.8293–0.9091 for estrogen…
Source →Near-infrared spectroscopy for urine-based cancer detection: a systematic characterization of the analytical challenges of a complex biofluid matrix.
Latysheva A, et al
A new systematic study of 85 urine samples reveals that near-infrared spectroscopy (NIRS) can detect kidney and prostate cancer signals in urine, but these signals are extraordinarily faint — cancer-discriminative information accounts for only 0.004% of total spectral variance for kidney cancer and …
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.