Deep learning based two-way feature depiction model for brain tumor detection

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: Overall accuracy: 96.25%; Recall: 96.34%; Precision: 96.31%; F1-score: 96.32%

• Primary Outcome: Four-class brain tumor classification accuracy on Brain MRI dataset using the TWFDM model

• Nnt Or Sensitivity: Sensitivity (recall) reported as 96.34% aggregate across four classes; class-specific sensitivity, specificity, and likelihood ratios not reported

• Confidence Interval: Not reported

Clinical Application

Technical feasibility as a computational pipeline is demonstrated, but clinical feasibility is unestablished. Integration into PACS or radiology reporting workflows, regulatory approval, prospective clinical validation, and radiologist acceptance testing have not been addressed. Computational requirements for IPSO-based feature selection may present deployment challenges in resource-limited settings. In Australia, brain tumour diagnosis follows RACGP and Cancer Australia guidelines, with MRI as the primary imaging modality and histopathological confirmation via biopsy or surgical resection as the definitive reference standard. The TGA would require prospective clinical validation data, demonstrated safety, and regulatory submission as a Software as a Medical Device (SaMD) under the TGA's Digital Health framework before any such tool could be used clinically. The model is not PBS-listed and has no current pathway to reimbursement. Australian neuro-oncology centres (e.g., through the Australian Brain Tumour Collaborative) would require multi-site validation across diverse MRI platforms before adoption. The study does not address Australian-specific epidemiological data, where approximately 1,900 Australians are diagnosed with brain tumours annually (Cancer Australia). Theoretically applicable to patients undergoing brain MRI for suspected glioma or other primary brain tumours; however, the study population is undefined and the model has not been validated in any clinical cohort

Abstract

Brain tumors are one of the most fatal disorders that cause one of the highest mortalities in the world. Gliomas are the most common primary brain tumors originating from glial cells in the central nervous system. Traditionally, a tissue sample is extracted and examined for its genetic and characteristic properties. This method is invasive, painful, and takes a longer period to produce results. Various automatic Deep learning (DL) based schemes have been presented for the brain glioma detection, but they lack due to poor explainability, lower generalization, poor feature depiction, class imbalance problem and lower detection rate. This paper presents a deep learning based brain tumor detection using two way feature depiction model (TWFDM) that combines the 2D-Deep Convolution Neural Network (DCNN) and 1D-DCNN. The 2D-DCNN accepts the raw MRI images and the 1D-DCNN accepts the handcrafted local binary pattern (LBP), gray level cooccurrence matrix (GLCM), and Histogram of Oriented Gradient (HOG) features. Furthermore, improved particle swarm optimization (IPSO) is used for feature selection to minimize the computational complexity of the TWFDM system. The proposed TWFDM achieves an overall accuracy of 96.25%, a recall of 96.34%, a precision of 96.31%, and an F1-score of 96.32% on the Brain MRI dataset for four-class classification, representing an important improvement over traditional techniques.

References

1. 1.Urooj, S., Napte, K., & Alsubaie, N. (2026). Deep learning based two-way feature depiction model for brain tumor detection. PLoS ONE. https://doi.org/10.1371/journal.pone.0344291