Automated polyp measurement comparable to manual method
The size of a polyp detected during a CT colonography procedure is of primary importance in diagnosis and treatment decisions due to the correlation between larger-size polyps and the risk of malignancy. Thus, a reliable measurement technique is a crucial tool for the interpreting physician. A research team from the Netherlands has developed a protrusion method for the automated estimation of polyp size on CT colonography that may contribute to a practical evaluation strategy for these exams.
“A clear advantage of the automated method is that it does not suffer from intraobserver variation,” wrote the authors of a multi-institution study published in this month’s American Journal of Roentgenology. “Moreover, the automated method can be calibrated once, whereas each observer may need individual calibration. Therefore, automated size measurement may well contribute to a practical evaluation strategy.”
Researchers from the departments of radiology, medical physics, gastroenterology and hepatology at Academic Medical Center in Amsterdam, along with colleagues from the department of imaging science and technology at Delft University of Technology, studied the accuracy and measurement variability of an automated measurement technique under varying scanning conditions using both phantom and patient data. The performance of the algorithm was compared with that of both 2D and 3D manual measurements by two clinicians experienced in evaluating CT colonography.
The phantom contained 15 asymmetric plasticine objects and 7 hemispherical acrylic resin objects. Scans of 10 patients were selected from an ongoing CT colonography trial at the institutions, which included patients at increase risk of colorectal cancer. CT images of the phantom and patients were acquired from a Philips Medical Systems’ Brilliance 64-slice CT system.
Automated polyp measurement was implemented on a proprietary experimental version of the Philips’ ViewForum 6.1 workstation. The scientists’ automated protrusion method entails estimation of the colonic wall needed to digitally remove a presume lesion.
“For example, iteratively moving the points on the convex parts of the polyp (i.e., the protruding part) inward effectively flattens the object,” the authors wrote. “After a certain amount of deformation, the surface flattening is such that the protrusion is removed. Thus the surface looks as if the object were never there. The amount of displacement is a measure of protrusion. The polyp is delimited by a threshold of the deformation field. The size is obtained by fitting an ellipse. The polyp measurement is the largest diameter of the ellipse.”
The researchers noted that the measurement variability of the automated method was significantly less than that of all observer scenarios (2D and 3D, both phantom and patient), with the exception of 3D measurements by one of the two interpreting clinicians. They also found that the interobserver variability of 2D measurements by the clinicians was significantly greater than the variability of 3D measurement.
“The systematic difference between the observers regarding 2D and 3D measurement signifies that separate correction values may be needed,” they wrote. “In other words, it may indicate that observers need to be calibrated individually to avoid systematic error.”
The automated method developed by the scientists, in contrast to manual measurement techniques currently deployed, merely requires user interaction to indicate a specific protrusion.
“Our findings indicated that there is reduced variability in polyp size with automated measurement of phantoms,” the researchers wrote. “The variability of automated measurement is in the same range as that of manual measurement of polyps from patients.”
“A clear advantage of the automated method is that it does not suffer from intraobserver variation,” wrote the authors of a multi-institution study published in this month’s American Journal of Roentgenology. “Moreover, the automated method can be calibrated once, whereas each observer may need individual calibration. Therefore, automated size measurement may well contribute to a practical evaluation strategy.”
Researchers from the departments of radiology, medical physics, gastroenterology and hepatology at Academic Medical Center in Amsterdam, along with colleagues from the department of imaging science and technology at Delft University of Technology, studied the accuracy and measurement variability of an automated measurement technique under varying scanning conditions using both phantom and patient data. The performance of the algorithm was compared with that of both 2D and 3D manual measurements by two clinicians experienced in evaluating CT colonography.
The phantom contained 15 asymmetric plasticine objects and 7 hemispherical acrylic resin objects. Scans of 10 patients were selected from an ongoing CT colonography trial at the institutions, which included patients at increase risk of colorectal cancer. CT images of the phantom and patients were acquired from a Philips Medical Systems’ Brilliance 64-slice CT system.
 | |
| Phantom colon with plasticine objects. Photograph shows two halves of cylinder inserted into tight-fit second cylinder before scanning. Image and caption by permission of the American Roentgen Ray Society. |
“For example, iteratively moving the points on the convex parts of the polyp (i.e., the protruding part) inward effectively flattens the object,” the authors wrote. “After a certain amount of deformation, the surface flattening is such that the protrusion is removed. Thus the surface looks as if the object were never there. The amount of displacement is a measure of protrusion. The polyp is delimited by a threshold of the deformation field. The size is obtained by fitting an ellipse. The polyp measurement is the largest diameter of the ellipse.”
The researchers noted that the measurement variability of the automated method was significantly less than that of all observer scenarios (2D and 3D, both phantom and patient), with the exception of 3D measurements by one of the two interpreting clinicians. They also found that the interobserver variability of 2D measurements by the clinicians was significantly greater than the variability of 3D measurement.
“The systematic difference between the observers regarding 2D and 3D measurement signifies that separate correction values may be needed,” they wrote. “In other words, it may indicate that observers need to be calibrated individually to avoid systematic error.”
The automated method developed by the scientists, in contrast to manual measurement techniques currently deployed, merely requires user interaction to indicate a specific protrusion.
“Our findings indicated that there is reduced variability in polyp size with automated measurement of phantoms,” the researchers wrote. “The variability of automated measurement is in the same range as that of manual measurement of polyps from patients.”