
Prof. Muhammad Aamir
Sichuan Technology and Business University, China
Title: From MRI to Diagnosis: Deep Learning for Brain Tumor Classification
Abstract:
Medical imaging is now the largest source of data in healthcare, and every image must be read by a trained specialist. Imaging volume grows each year; the number of specialists does not. This gap is structural, and it appears in the same form across the field: a difficult visual judgment, made under time pressure, by an expert who is scarce. The same pattern underlies diabetic retinopathy screening, chest X-ray triage, histopathology, dermatology and neuro-radiology, which is why deep learning has been applied to all of them with broadly similar architectures.
This work examines that pattern through one representative problem: the classification of brain tumors from MRI. A four-stage method is presented, in which the scan is first enhanced to correct low contrast and uneven illumination; features are extracted by two pre-trained networks and combined into a single representation using partial least squares; candidate tumor regions are generated by agglomerative grouping and ranked by an edge-based score; and the aligned regions are classified as glioma, meningioma or pituitary. On a public benchmark of 3,064 contrast-enhanced scans, the method reaches 98.95% accuracy and 95.4% detection recall using only 100 proposals per image.
The feature-fusion result generalizes beyond this task: no single architecture captures all discriminative information, and two independent representations of the same input reinforce one another. The work closes by setting out what is still required before models of this kind can be used clinically; patient-level validation, external evaluation across sites, and prospective testing alongside clinicians.
Biography:
Muhammad Aamir received his Bachelor of Engineering degree in Computer Systems Engineering from Mehran University of Engineering & Technology, Jamshoro, Sindh, Pakistan, in 2008. He went on to earn a Master of Engineering degree in Software Engineering from Chongqing University, China, in 2014, and later completed his PhD in Computer Science and Technology from Sichuan University, Chengdu, China, in 2019. Currently, he is an Associate Professor in the College of Computer Science at Huanggang Normal University, China.
With extensive research experience and a prolific publication record, Dr. Aamir has made significant contributions to various fields including pattern recognition, computer vision, medical image processing, deep learning, and machine vision. His expertise and dedication to advancing these areas have earned him recognition as a peer reviewer for numerous prestigious journals and as an organizing committee member for several international conferences.