AI and Cancer Research ── Cancer biology ── 2026-10-25
Cancer molecular biology and AI
- Abstract
Diagnostic imaging represents one of the most promising clinical applications of artificial intelligence (AI) algorithms. The purpose of this study was to evaluate the customised convolutional neural network (CNN) SepNetDense for distinguishing between normal and abnormal canine brain magnetic resonance images (MRIs), with the aim of enhancing diagnostic efficiency and assisting radiologists in identifying abnormal images. The dataset comprised T1-weighted (T1w) pre- and post-contrast sequences in transverse, sagittal, and dorsal planes from 550 dogs, collected from four universities. Dogs were included if they had a complete clinical diagnosis confirmed either through histopathology or in accordance with current clinical consensus. Patients were randomly divided into a training set (n = 444), a validation set (n = 53), and a test set (n = 53). Each MRI was labelled on a slice-by-slice basis as normal or abnormal. The model was trained on 205 normal imaging datasets (e.g., extracranial aetiologies, idiopathic epilepsy, paroxysmal dyskinesia) and 239 abnormal ones (e.g., neoplasms, inflammatory lesions, other pathologies). The model correctly predicted 74% of the true normal slices in the test set as normal and 73% of the true abnormal as abnormal. A ROC analysis of the model’s prediction at the patient level revealed that, at a threshold of 51% abnormal slices per patient, the model reached an optimal balance of 83% sensitivity (95% CI 67–91%) and 78% (95% CI 54–91%) specificity, with a balanced accuracy of 80% (95% CI 69–92%). ANCOVA revealed that the CNN’s classification performance was influenced by multiple biological, technical, and institutional factors. Lesion diagnosis, institutional setting, and breed size had significant large effects, whereas body weight showed a significant medium effect. Significant interactions with large effect sizes were also observed between diagnosis and institute, as well as between breed size and weight. Additionally, a distinct pattern was observed in the distribution of prediction categories along the anatomical axes, with a trend towards better CNN performance in the central quartiles across all MRI sequences. Overall, the results demonstrate the feasibility of CNN-assisted triage of canine brain MRI studies and highlight diagnosis, institutional setting, breed size, and body weight as key determinants of model performance, emphasising the importance of heterogeneous datasets for the development of robust veterinary AI applications.
Journal IF-equivalent: 5.2 (OpenAlex 2-year mean citedness, value as of 2026-10-04, retrieved 2026-10-05; not the official Clarivate IF)Reference: Abani S, Laue M, Dickinson PJ, De Decker S, Gutierrez-Quintana R, Ghiasvand S, et al. CNN model-based image classification for canine brain MRI abnormalities. Sci Rep. 2026 Oct 23;16(1):33913. doi:10.1038/s41598-026-71470-w.Checked: Abstract only - Abstract
Two of the most severe forms of neurological issues, which can be treated only when diagnosed with a high degree of accuracy with the help of MRI, are brain tumors and Alzheimer disease (AD).The project under analysis considers the datasets of MRI that are not balanced and refers to the Synthetic Minority Oversampling Technique (SMOTE) to improve the classification process.They included several ML and DL models including the RF, SVM, Voting Classifier, CNN, VGG16, ResNet50, and Inception V3, as well as the AlexNet with LSTM, Xception, and the new Transformer and LSTM Enhanced Attention-Based Explainable CNN (TLEABLCNN).TLEABLCNN model consistently performed better than its baseline models with a classification accuracy of 96.1% in Scenario A, 97.4% in Scenario B and 98.2% in Scenario C. It also possessed superior macro-precision, recall and F1-scores.To detect abnormalities in the brain; the object recognition models of YOLOv5, YOLOv8, YOLOv9 and YOLOv11 were employed.YOLOv9 got the highest best mean average precision (mAP) of 91.5%.YOLOv8 achieved the best accuracy of 91.8%.Moreover, XAI techniques, such as heatmaps, were employed to generate the heatmaps that revealed the most significant portions of MRI scans that influenced model predictions.This simplified the results and enhanced professional trust.These findings indicate the strength of the framework in the discovery of brain disorders which can be explained.
Journal IF-equivalent: 0.0 (OpenAlex 2-year mean citedness, value as of 2026-10-07, retrieved 2026-10-08; not the official Clarivate IF)Reference: [No authors listed]. TLEABLCNN: Brain and Alzheimer’s Disease Detection Using Attention-Based Explainable Deep Learning and SMOTE Using Imbalanced Brain MRI. IJDDT. 2026 Oct 22;16(76s). doi:10.25258/ijddt.16.76s.161.Checked: Abstract only - Abstract
Super-resolution (SR) facilitates high-fidelity mapping beyond the spatial resolution capacity of clinical neuro-oncology MRI. While state-of-the-art deep learning algorithms have augmented this domain, such methods exclusively learn from spatial or spectral domains and are often compounded by tumor boundary over-smoothing and geometrically inconsistent pathologies. In this paper, we introduce Hy-DTGL, a hybrid-domain transformer with a global–local attention modal super-resolution (SR) framework that learns a joint reconstruction across four inter-dependent knowledge domains: local spatial, global spatial, spectral and wavelet. Using windowed self-attention, Hy-DTGL learns highly localized pathologic textures; pooled global attention constrains anatomical plausibility; Fourier-learnable modulation provides spectral energy per semblance; and multi-scale discrete wavelet tiling preserves multi-scale tumor morphology. Optimization occurs via a hybrid objective, combining Charbonnier loss, gradient edge-loss, and spectral domain loss. The proposed framework was trained on BraTS 2019 (BT_19) and evaluated on BT_19, BraTS 2020 (BT_20), BraTS 2023 (BT_23), and IXI under both in-domain and strict zero-shot settings, and achieves competitive performance relative to the evaluated baselines under the adopted degradation and evaluation protocol at ×2, ×4, and ×8 SR factors. In the BraTS 2020 ×4 test resolution, 37.46 dB PSNR and 0.986 SSIM are attained, outperforming deep learning CNN-based SR (ECFNet) by 2.47 dB and transformer-based (WavTrans) by 4.05 dB in PSNR. Ablative studies reveal the spatial, spectral and multi-scale constraints notably converge to a faithfully mapped image space with restored tumor boundary delineation and global-structure consistency. We also show the Hy-DTGL generalizability by how it improves downstream tumor segmentation metrics: the super-resolved scans improve the segmentation Dice coefficient by 0.063 and significant reductions are observed in the Hausdorff Distance from 19.05 mm to 13.01 mm relative to the original low-resolution images, suggesting potential utility for downstream neuro-oncological image analysis.
Journal IF-equivalent: 3.6 (OpenAlex 2-year mean citedness, value as of 2026-10-23, retrieved 2026-10-24; not the official Clarivate IF)Reference: Hassan M, Alshdaifat N, Owida HA, Hamadneh T, Mashagba HA, Aziz ABA, et al. Integrating Spatial, Spectral, and Wavelet Domains via Global–Local Attention for Robust Multi-Modal MRI Super-Resolution. Algorithms. 2026 Oct 21;19(10):904. doi:10.3390/a19100904.Checked: Abstract only - Abstract
Major histocompatibility complex class I (MHC I) molecules initiate CD8 + T cell immune responses by presenting antigenic peptides, a process fundamentally governed by the spatial interaction network mediated by multi-residue synergy between the binding groove and the peptide. Accurate modeling of this mechanism is crucial for tumor neoantigen screening and personalized immunotherapy. However, existing methods are largely confined to sequence-level modeling, failing to capture cross-residue higher-order structural dependencies and fine-grained binding interface interactions. To address these challenges, we propose pMHCBind, a prediction model that integrates a protein language model with a hypergraph neural network. The model employs ESM-MSA-1b and ESM-2 to extract evolution-aware residue-level embeddings for MHC I molecules and antigenic peptides, respectively, and constructs a peptide-specific hypergraph based on spatial contact relationships to model higher-order interactions via hypergraph convolution. Subsequently, cross-molecular feature fusion is employed to characterize the interaction patterns on the binding interface for presentation ability prediction. In five-fold cross-validation on the NetMHCpan-4.1 benchmark dataset and the test performance on the Anthem independent test dataset, pMHCBind consistently outperforms a variety of state-of-the-art methods across multiple metrics, including the area under receiver operating characteristic curve (AUC) and area under the precision-recall curve (AUPR). Furthermore, interpretability analysis reveals that the model effectively identifies key interaction modes, such as hydrogen bonds, salt bridges, and hydrophobic interactions, between the binding groove and the peptide, demonstrating strong structural interpretability. Collectively, pMHCBind provides a promising and biologically interpretable computational tool for neoantigen recognition in precision cancer immunotherapy. 主要组织相容性复合体I类分子(major histocompatibility complex class I, MHC I)通过呈递抗原肽激活CD8 + T细胞免疫应答，其结合槽与抗原肽之间的多位点残基协同互作决定了抗原呈递的特异性。准确刻画这一互作过程对于肿瘤新抗原筛选和个体化免疫治疗至关重要。然而，现有预测方法多基于序列信息建模，难以充分表征结合过程中复杂的高阶结构依赖及跨分子残基互作。为此，本文提出融合蛋白质大语言模型与超图神经网络的MHC I-抗原肽互作预测模型pMHCBind。该模型利用ESM-MSA-1b与ESM-2分别提取MHC分子与抗原肽的残基级表示，并基于空间接触关系构建抗原肽超图，通过超图卷积建模残基间高阶关联;随后融合MHC与抗原肽特征，刻画结合界面的互作模式，实现抗原呈递能力预测。在NetMHCpan-4.1基准数据集的五折交叉验证以及Anthem独立测试集上的实验结果表明，pMHCBind在受试者工作特征曲线下面积(area under the receiver operating characteristic curve, AUC)、精确率-召回率曲线下面积(area under the precision-recall curve, AUPR)等评价指标上均优于多种主流方法。可解释性分析进一步表明，该模型能够有效识别结合槽与抗原肽之间的氢键、盐桥及疏水相互作用等关键结合模式，展现出良好的结构解析能力。综上，pMHCBind能够兼顾预测性能与生物学可解释性，可为精准肿瘤免疫治疗中的新抗原筛选提供有效的计算工具。.
Journal IF-equivalent: 0.0 (OpenAlex 2-year mean citedness, value as of 2026-10-24, retrieved 2026-10-25; not the official Clarivate IF)Reference: [No authors listed]. [pMHCBind: predicting major histocompatibility complex class I-peptide interactions via protein large language models and hypergraph neural networks]. Sheng wu gong cheng xue bao = Chinese journal of biotechnology. [Epub ahead of print]. doi:10.13345/j.cjb.260422.Checked: Abstract only