Consistent surgeon evaluations of three-dimensional rendering of PET/CT scans of the abdomen of a patient with a ductal pancreatic mass

Abstract

Two-dimensional (2D) positron emission tomography (PET) and computed tomography (CT) are used for diagnosis and evaluation of cancer patients, requiring surgeons to look through multiple planar images to comprehend the tumor and surrounding tissues. We hypothesized that experienced surgeons would consistently evaluate three-dimensional (3D) presentation of CT images overlaid with PET images when preparing for a procedure. We recruited six Jefferson surgeons to evaluate the accuracy, usefulness, and applicability of 3D renderings of the organs surrounding a malignant pancreas prior to surgery. PET/CT and contrast-enhanced CT abdominal scans of a patient with a ductal pancreatic mass were segmented into 3D surface renderings, followed by co-registration. Version A used only the PET/CT image, while version B used the contrast-enhanced CT scans co-registered with the PET images. The six surgeons answered 15 questions covering a) the ease of use and accuracy of models, b) how these models, with/without PET, changed their understanding of the tumor, and c) what are the best applications of the 3D visualization, on a scale of 1 to 5. The six evaluations revealed a statistically significant improvement from version A (score 3.6±0.5) to version B (score 4.4±0.4). A paired-samples t-test yielded t(14) = -8.964, p<0.001. Across the surgeon cohort, contrast-enhanced CT fused with PET provided a more lifelike presentation than standard CT, increasing the usefulness of the presentation. The experienced surgeons consistently reported positive reactions to 3D surface renderings of fused PET and contrast-enhanced CT scans of a pancreatic cancer and surrounding organs. Thus, the 3D presentation could be a useful preparative tool for surgeons prior to making the first incision. This result supports proceeding to a larger surgeon cohort, viewing prospective 3D images from multiple types of cancer.

Figure 1. 2D CT/PET fusion image slice of an anonymized patient with a ductal pancreatic mass.

Regions with the highest 2′-[¹⁸F]fluorodeoxyglucose emission are colored red here while the lowest emissions are colored blue. A surgeon currently looks back and forth through a stack of such images to gain an understanding of anatomy surrounding the lesion. For this image in the coronal plane, the displayed PET window was narrowed to accentuate the location of high uptake in the pancreas as well as another hotspot in the liver.

Figure 2. 3D rendering of an anonymized patient’s abdomen with a ductal pancreatic mass, version A.

Organs displayed in this rendering include a) rib cage, b) liver, c) intestines, d) stomach, e) pancreas, and f) aorta.

Figure 3. 3D rendering of the abdomen with organs stripped away to display the pancreatic tumor.

The liver, stomach, and intestines were not visualized for a clearer view of the pancreas and its surroundings. Organs displayed in this rendering include a) rib cage, b) spleen, c) pancreas, d) duodenum, e) right kidney, f) left kidney, g) aorta, h) vena cava, i) superior mesenteric artery, j) adrenal gland, and k) pancreatic tumor.

Figure 4. 3D rendering of an anonymized patient’s abdomen with a ductal pancreatic mass, version B.

Organs displayed in this rendering include a) rib cage, b) spine, c) liver, d) intestines, e) stomach, f) right gastroepiploic vein, and g) superior mesenteric vein, h) superior mesenteric artery, i) left common iliac artery, and j) right common iliac artery.

Figure 5. 3D rendering of the abdomen with organs stripped away to display the pancreatic tumor.

The liver, stomach, and intestines were not visualized for a clearer view of the pancreas. Organs displayed in this rendering include a) rib cage, b) spine, c) spleen, d) pancreas, e) duodenum, f) right kidney, g), left kidney h) aorta, i) vena cava, j) portal vein, k) right gastroepiploic vein, l) superior mesenteric vein, m) celiac artery, n) superior mesenteric and intestinal arteries o) left common iliac artery, p) right common iliac artery, and q) pancreatic tumor.

Figure 6. 3D rendering of the ductal pancreatic mass, version A, fused with PET data.

A transparent overlay of (a) the pancreas over (b) the surface rendering of the high ¹⁸FDG uptake region of the pancreatic tumor (orange).

Figure 7. Average surgeon evaluation scores for version A and version B of the 3D visualization.

Average scores, with error bars, are shown for questions on a) the ease of use and accuracy of models, b) how these models, with/without PET, changed their understanding of the tumor, and c) what are the best applications of the 3D visualization.

Figure 8. Score distribution on the ease of use and accuracy of models.

Percentages of surgeons who assigned a particular score for both Version A (solid colors with hatch marks) and Version B (solid colors) [Q1: The display hardware is easy/comfortable to use, Q2: I found it easy to manipulate/re-position the image, Q3: The organs/structures are accurately represented (accuracy of segmentation), Q4: Colors/textures are appropriate (accuracy of rendering), Q5: I am satisfied with the level of detail that is presented, Q6: The model provides me with adequate reference to surrounding structures, Q7: The overall 3D image appears realistic (matches what I expect to see in the OR)].

Figure 9. Score distribution on how these models, with/without PET, changed their understanding of the tumor.

Percentages of surgeons who assigned a particular score for both Version A (solid colors with hatch marks) and Version B (solid colors) [Q8: By simply viewing the image(s) in the 3D model, I get a better understanding of the tumor and its relationship to the surrounding organs, Q9: By simply viewing the image(s) in the 3D model, my plan for how to approach this tumor changed (as compared to traditional CT images), Q10: By overlaying the PET data in the 3D model, I get a better understanding of the tumor and its relationship to the surrounding organs, Q11: By overlaying the PET data in the 3D model, my plan for how to approach this tumor changed (as compared to traditional CT images)].

Figure 10. Score distribution on what are the best applications of the 3D visualization.

Percentages of surgeons who assigned a particular score for both Version A (solid colors with hatch marks) and Version B (solid colors) [Q12: I would want to use this 3D image to plan an operation for a patient with a specific tumor, Q13: I would want to use this 3D image with PET overlay to plan an operation for a patient with a specific tumor, Q14: I would want to have this system available to me in the OR, for the reference during an actual operation, Q15: I believe that this system would help residents/assistant better prepare for the operation].