CHICAGO — Automated image analysis is gaining ground in the field, but the pathologist still has the advantage, say researchers here at the College of American Pathologists (CAP) 2014 meeting.
While evaluation of automated analysis is still in the early stages, Maryam Abdelghani, MD, from Emory University Hospital in Atlanta, Georgia, explained to Medscape Medical News, "in the long run, it could help with time management."
Dr. Abdelghani presented her research at a poster session here at the meeting. Her abstract is representative of an increasing number of studies exploring the use of automated image analysis.
Her team evaluated tissue microarrays from neoplastic and nonneoplastic liver (n = 119 cases). Two pathologists calculated a Q score for each specimen (intensity times percentage of tumor cells staining) using immunohistochemistry. They scored the samples as 0 (negative), 1, 2, or 3 (strong). Computer-assisted image analysis was performed using the Aperio system, and the result was a quantification of the immunostain.
The investigators compared visual vs image analytic quantitation methods of immunostains. They also examined the correlation between p62 and ubiquitin staining and outcome in hepatocellular carcinoma.
The investigators found that high Q scores were associated with high Aperio scores for both p62 and ubiquitin for all comparisons except for nuclear p62. They also identified a trend toward worse survival with higher Aperio scores.
Table. Correlation Between Q Score and Aperio in Liver Neoplastic Tissue
|Stain||Cellular||Correlation of Q Score With Aperio||P Value|
"More and more, we want to try and assess things as objectively as possible," Philip Cagle, MD, editor-in-chief of Archives of Pathology & Laboratory Medicine, explained in an interview with Medscape Medical News. "When you can quantify things using an instrument, you can get a much more objective number."
While acknowledging that this is an interesting and growing field, Dr. Abdelghani said that at this point "there are not enough data to say automated data analysis can replace a pathologist."
She points out that pathologists have the unique ability to determine whether the staining is cytoplasmic, nuclear, or membranous. At this point, the automatic image analysis cannot do this.
Dr. Abdelghani and Dr. Cagle have disclosed no relevant financial relationships.
College of American Pathologists (CAP) 2014. Abstract 54. Presented September 8, 2014.
Medscape Medical News © 2014 WebMD, LLC
Send comments and news tips to firstname.lastname@example.org.
Cite this: Automated Image Analysis Cannot Replace Pathologist - Medscape - Sep 12, 2014.