6737

Deep Learning Convolutional Neural Network (CNN) for Automatic Detection and Diagnosis of Sacroiliitis in CT Scans

Joskowicz Leo, HUJI, School of Computer Science and Engineering, Computer Science

Keywords

Sacroiliitis, Sacroiliac joint, Diagnosis, Medical Image Processing, Lower back pain

Current development stage

TRL3 Experimental proof of concept

Application

  • Diagnosis of Sacroiliitis and separation from other types of lower back pain and associated maladies is non-trivial and results, in some cases, in late diagnosis.
  • Numerous CT scans are performed for patients with non-specific lower back pain while the condition goes undiagnosed.
  • Early diagnosis will enable preventive treatment, and reduce costs associated with repeat CT scans and ongoing, ineffective, treatment.

Our Innovation

The research group developed a deep learning convolutional neural network (CNN) for automatic detection and diagnosis of Sacroiliitis in CT scans. The CNN automatically identifies both Sacroliac joints then proceeds to grade the level of Sacroiliitis, in each joint, based on an averaging of grading in each CT slice.

The Advantages 

  • Fully automatic
  • Very high sensitivity
  • Grading system based on individual grading per CT slice.

Opportunity

Close to 30% of the population reports lower back pain, many of which are sent for a CT scan for diagnosis

The technology taps this market by addressing this frequently undiagnosed condition. The use of the deep learning CNN will not be limited to patients presenting with Sacroiliitis but potentially to every CT scan performed for patients with an undiagnosed lower back pain.

Contact for more information:

Mel Larrosa
VP Business Development Healthcare
+972-2-6586692
Contact ME: