Abstract
Background: Contrast enhanced mammography (CEM) is a functional imaging technique with similar accuracy to MRI. Previous work suggests CEM quantitative temporal enhancement values using both freehand-ROIs and oval-ROIs follow a similar pattern to MRI time-intensity curves. We investigate whether subtracting background enhancement, a pre-enhancement image substitute, improves accuracy.
Methods: Enhancing mass-lesions on CEM with contemporaneous MRI studies were included. Two MLO views were acquired, 3 and 9-minutes post-contrast administration. Segmentation of background-ROIs and CEM lesions (oval and freehand-ROIs) were performed on both views by a radiologist blinded to MRI. Lesion enhancement was subtracted from background enhancement, the mean, 90th and 99th centile greyscale values (GSV) recorded, and temporal change calculated. Differences between CEM temporal enhancement according to MRI-curve type were calculated using a Mann Whitney U test.
Results: Of 55 lesions segmented, 19 produced type-1 curves and 35 type-3 curves. A solitary type-2 curve lesion was excluded. With respect to oval-ROIs, lesions with MRI type-1 curves demonstrated increasing CEM mean, 90th and 99th centile GSVs, lesions with type 3 curves demonstrated decreasing mean, CEM 90th and 99th centile GSVs with significant variation between MRI cohorts, p<0.05. However only mean and 90th centile GSVs demonstrated significant differences with the freehand-ROIs. Greatest variation was seen for oval-ROI 90th centile: temporal GSV 1.53 vs -8.90, p=0.003 and oval-ROI 99th centile: temporal GSV 0.68 vs 1.53, p=0.005, for type-1 and type-3 curves respectively.
Discussion: Significant differences in CEM temporal GSV are demonstrated between MRI curve types. Accuracy is not improved by subtracting background enhancement
Methods: Enhancing mass-lesions on CEM with contemporaneous MRI studies were included. Two MLO views were acquired, 3 and 9-minutes post-contrast administration. Segmentation of background-ROIs and CEM lesions (oval and freehand-ROIs) were performed on both views by a radiologist blinded to MRI. Lesion enhancement was subtracted from background enhancement, the mean, 90th and 99th centile greyscale values (GSV) recorded, and temporal change calculated. Differences between CEM temporal enhancement according to MRI-curve type were calculated using a Mann Whitney U test.
Results: Of 55 lesions segmented, 19 produced type-1 curves and 35 type-3 curves. A solitary type-2 curve lesion was excluded. With respect to oval-ROIs, lesions with MRI type-1 curves demonstrated increasing CEM mean, 90th and 99th centile GSVs, lesions with type 3 curves demonstrated decreasing mean, CEM 90th and 99th centile GSVs with significant variation between MRI cohorts, p<0.05. However only mean and 90th centile GSVs demonstrated significant differences with the freehand-ROIs. Greatest variation was seen for oval-ROI 90th centile: temporal GSV 1.53 vs -8.90, p=0.003 and oval-ROI 99th centile: temporal GSV 0.68 vs 1.53, p=0.005, for type-1 and type-3 curves respectively.
Discussion: Significant differences in CEM temporal GSV are demonstrated between MRI curve types. Accuracy is not improved by subtracting background enhancement
Original language | English |
---|---|
Publication status | Published - Jun 2024 |
Event | SINAPSE 2024 ASM - University of Stirling, Stirling, United Kingdom Duration: 12 Jun 2024 → 12 Jun 2024 https://www.sinapse.ac.uk/events/2024-sinapse-asm/ |
Conference
Conference | SINAPSE 2024 ASM |
---|---|
Country/Territory | United Kingdom |
City | Stirling |
Period | 12/06/24 → 12/06/24 |
Internet address |