Computerized interactive morphometry. An expert system for the diagnosis of lymphoid-rich effusions

Area of Science:

• Cytopathology
• Medical Informatics
• Hematology

Background:

• Distinguishing malignant lymphoma from benign lymphocytosis in effusions is clinically significant.
• Traditional methods can be labor-intensive and require specialized expertise.
• Computerized analysis offers potential for objective and efficient diagnostic support.

Purpose of the Study:

• To evaluate the efficacy of a microcomputer-based video system for computerized interactive morphometry (CIM) in diagnosing lymphoid-rich effusions.
• To assess CIM's accuracy in differentiating malignant lymphoma from benign lymphocytosis.
• To determine CIM's suitability for diagnosing chronic lymphocytic leukemia (CLL) in effusions.

Main Methods:

• A Papanicolaou-fixed and stained cytospin smear system was developed for lymphoid cell analysis.
• Lymphoid cells were selected from real-time images of pleural, peritoneal, or pericardial effusions.
• An expert system classified cells based on nuclear profile size and shape, differentiating benign from malignant.

Main Results:

• The CIM system correctly classified 91 out of 104 effusions (87.5%).
• Excluding chronic lymphocytic leukemia (CLL) cases, CIM achieved 96.5% predictive value for malignant lymphoma and 94.0% for benign lymphocytosis.
• CIM consistently misclassified eight CLL effusions as benign, indicating unsuitability for this specific diagnosis.

Conclusions:

• Computerized interactive morphometry (CIM) shows promise for diagnosing lymphoid effusions, with high accuracy in differentiating malignant lymphoma from benign lymphocytosis.
• CIM is not suitable for diagnosing malignancy in effusions from patients with chronic lymphocytic leukemia (CLL).
• Further refinement of CIM may be needed for specific hematologic conditions like CLL.