Siemens (Israel)

Nuclear magnetic resonance is one of the cornerstones of modern medicine and biomedical research. Over the past several decades, the speed and precision of in-vivo magnetic resonance imaging (MRI) and spectroscopy (MRS) have increased by leaps and bounds, by utilizing sophisticated excitation and acquisition techniques, from parallel imaging and compressed sensing to magnetic resonance fingerprinting. However, these approaches have all been static in nature, fixing measurement parameters in advance, in anticipation of a wide range of expected tissue parameter values, and are therefore sub-optimal for any given subject. We depart from the conventional framework of magnetic resonance and propose a new approach - termed adaptive magnetic resonance - which binds acquisition and excitation, by using the measured signal to update and fine-tune the measurement parameters in real time. This targets the specific tissue characteristics of the subject while they are being scanned. Adaptive magnetic resonance provides a completely new and previously-untapped avenue for improving the sensitivity and specificity of in-vivo magnetic resonance across all tissue contrast mechanisms. Equivalently, it can accelerate data acquisition compared to non-adaptive schemes, by obtaining the same precision using fewer, optimally tuned excitations. We demonstrate that an adaptive pulse sequence for measuring the transverse relaxation time (T2) of metabolites in-vivo improves upon the precision of static approaches by a factor of ≈ 1.7 - or, alternatively, accelerates acquisition 2.5-fold.

Computer aided detection (CAD) system for cancer detection from X-ray images is highly requested by radiologists. For CAD systems to be successful, a large amount of data has to be collected. This poses new challenges for developing learning algorithms that are efficient and scalable to large dataset sizes. One way to achieve this efficiency is by using good feature selection.

This paper deals with learning spiculation scores of masses in a supervised manner Three spiculation score prediction models treating the score either as a continuous or ordinary variable are presented These models were compared on a data-set of 255 masses.