<i>Radiology</i> editorial board highlights top research papers of 2020

Over 100 Total Lots Up For Auction at One Location - CA 04/30

Radiology editorial board highlights top research papers of 2020

Lauren Dubinsky, Senior Reporter | December 10, 2020
Artificial Intelligence CT MRI Rad Oncology Ultrasound X-Ray

This study shows that although neurologic symptoms are common in COVID patients, MR does not frequently show abnormalities. The causes of these abnormalities were not discovered and none of the patients had COVID in their cerebrospinal fluid.

“It’s hard to be sure which of these findings were related directly to COVID-19 and what is the result of just being sick in the ICU," said Dr. Chris Hess, chair of radiology at University of California, San Francisco (UCSF).

Servicing GE/Siemens Nuclear Medicine equipment with OEM trained engineers

Numed, a well established company in business since 1975 provides a wide range of service options including time & material service, PM only contracts, full service contracts, labor only contracts & system relocation. Call 800 96 Numed for more info.

Study: Artificial Intelligence Systems Approaching Neuroradiologist-level Differential Diagnosis Accuracy at Brain MRI

Researchers at UCSF identified 178 patients with 19 diseases ranging from tumors to vascular disease, and designed a composite AI system that models how a radiologist makes an imaging diagnosis.

“Instead of looking for a single diagnosis within an image, it uses AI to detect a broad class of lesions and then uses human domain knowledge to assign a differential diagnosis for detecting lesions," explained Hess.

They used Bayesian inference for this process, which is a way of embedding neuroradiologist knowledge to differentiate between the likelihood of different lesions, and found that the AI system included the correct diagnosis in 91% of the cases. In addition, they found that that the AI had better results than residents, general radiologists and one of two neuroradiology fellows, and was equivalent to the performance of the academic faculty.

“It was also noted that the removal of clinical features from the algorithm resulted in a decline in performance of the system by 20%," said Hess.

Study: Diagnostic Accuracy of CEUS LI-RADS for the Characterization of Liver Nodules 20 mm or Smaller in Patients at Risk for Hepatocellular Carcinoma

Contrast-enhanced ultrasound (CEUS) is now a first-line test for at-risk patients in many countries worldwide. In addition, LI-RADS offers the opportunity to improve diagnosis in at-risk patients, but performance still needs further evaluation, according to Dr. Vicky Goh, professor and chair of clinical cancer imaging at Kings College, London.

A retrospective single-center study recruited 172 at-risk patients with 175 treatment-naïve liver nodules that were 2 centimeters or smaller. CEUS LI-RADS and the World and European criteria (WFUMB-EFSUMB) for diagnosing HCC were compared against the composite reference standard of imaging follow-up and histology. Inter-reader agreement was assessed.

The data revealed that 60% of nodules were HCCs and that inter-reader agreement was excellent. The study also found that CEUS LI-RADS was highly specific and had a high positive predictive value of 98% when LR5 was used. However, the World and European criteria only had a 89% sensitivity rate and a 87% specificity rate.

You Must Be Logged In To Post A Comment