What Is IEEE Transactions on Medical Imaging?

IEEE Transactions on Medical Imaging (TMI) is a peer-reviewed journal co-published by the IEEE Engineering in Medicine and Biology Society and the IEEE Signal Processing Society that covers imaging of body structure, morphology, and function, including cell and molecular imaging and all forms of microscopy. The journal unifies the sciences of medicine, biology, and imaging by publishing work where instrumentation, hardware, software, mathematics, physics, and clinical biology interact through new analysis methods. Founded in 1982 under the editorship of Michel M. Ter-Pogossian, TMI launched as a quarterly outlet for reconstruction and processing work in modalities then emerging into clinical use, including computed tomography (CT), single-photon emission computed tomography (SPECT), positron emission tomography (PET), nuclear magnetic resonance imaging, and ultrasound. The journal has since expanded its scope to include virtually every modality used to image living tissue, and it is consistently among the most highly cited publications in biomedical engineering.

Image Acquisition and Reconstruction

A foundational sub-area of TMI is the physics and mathematics of how raw detector measurements are converted into images. Research in this sub-area addresses reconstruction algorithms for CT (including iterative and model-based approaches that reduce radiation dose compared to filtered backprojection), MRI pulse sequence design and parallel imaging acceleration techniques, PET and SPECT attenuation correction and statistical reconstruction, and ultrasound beamforming and synthetic aperture imaging. Papers examine how detector physics, measurement noise, and geometrical constraints impose resolution and contrast tradeoffs, and how those tradeoffs can be improved through computational methods applied either during acquisition or in post-processing. The journal has been a primary venue for reporting advances in compressed sensing applied to MRI, beginning around 2007, which demonstrated that images could be reconstructed from far fewer k-space measurements than the Nyquist criterion would suggest. The scope and submission priorities of TMI are detailed on the IEEE TMI website.

Image Processing, Segmentation, and Registration

TMI publishes extensively on the algorithms used to analyze medical images after acquisition: detecting and delineating anatomical structures, registering images from different time points or imaging modalities, and quantifying morphological or functional changes that indicate disease. Segmentation methods range from classical approaches based on deformable contours and atlas-based label propagation to deep learning architectures that learn to label pixels or voxels directly from annotated training sets. Registration work addresses the estimation of geometric transforms that align images acquired under different conditions, a problem central to longitudinal studies, image-guided surgery, and multi-modal fusion. Papers in this sub-area must address both methodological novelty and rigorous evaluation against clinical ground truth, a standard the journal enforces through its requirement for strong application evidence alongside algorithmic contribution. Full publication archives from 1982 to the present are accessible through IEEE Xplore's Transactions on Medical Imaging collection.

Machine Learning and Computer-Aided Diagnosis

The journal has increasingly published work on machine learning methods applied to medical imaging, including convolutional neural networks for pathology classification in histology slides, detection of nodules in chest CT, grading of diabetic retinopathy in fundus photographs, and segmentation of tumors in MRI. Research in this sub-area examines the statistical challenges specific to medical imaging, including small labeled training sets relative to image dimensionality, class imbalance between rare lesions and normal tissue, and distribution shift between training data collected at one institution and test data from another. Fairness and interpretability of automated diagnostic models have become active research directions, reflecting the regulatory and clinical stakes of deploying algorithmic decision support in patient care. Biomedical imaging informatics resources maintained by NIH's National Library of Medicine complement the methodological contributions reported in TMI.

Applications

IEEE Transactions on Medical Imaging covers research with applications across a wide range of clinical and research domains, including:

• Radiology and oncology aided by automated lesion detection and characterization
• Radiation therapy planning using multi-modal image segmentation
• Neuroscience and psychiatry research using functional and structural brain MRI
• Surgical navigation systems using intraoperative ultrasound and fluoroscopy
• Pathology analysis through whole-slide digital imaging and deep learning