Back to the list
Congress: ECR25
Poster Number: C-12160
Type: Poster: EPOS Radiologist (scientific)
Authorblock: K. B. Krug1, D. Schafigh1, A-I. Iuga1, M. Weisthoff1, D. Maintz1, P. A. Baltzer2, W. Malter1, M. Hellmich1, K. Klein1; 1Cologne/DE, 2Vienna/AT
Disclosures:
Kathrin Barbara Krug: Nothing to disclose
Darius Schafigh: Nothing to disclose
Andra-Iza Iuga: Nothing to disclose
Mathilda Weisthoff: Nothing to disclose
David Maintz: Nothing to disclose
Pascal At Baltzer: Nothing to disclose
Wolfram Malter: Nothing to disclose
Martin Hellmich: Nothing to disclose
Konstantin Klein: Nothing to disclose
Keywords: Breast, CT-Quantitative, Mammography, Decision analysis, Cancer, Neoplasia, Tissue characterisation
Results

During the survey period, 5.591 contrast-enhanced examinations of the chest were carried out in female patients aged ≥18 years on the DECT system. In 687 of the 5.591 CT-examinations (12.3%) displayed a breast lesion. No diagnosis was possible in 161 of the 687 DLCT-examinations (23.4%) due to lacking diagnosis confirmation and to combined residual cancer and postoperative hematoma in 2 cases (0.3%). 109 patients (15.9%) were excluded due to ongoing antiproliferative/antihormonal therapy.

The study group thus comprised 415 female patients with a mean age of 59.7 years (range 18–91 years) and 415 breast masses with a diagnostic assignment to benignity or malignancy confirmed by histology (155, 37.4%), a constant CT-course of ≥2 years without interfering therapy (202, 48.8%) and/or non-invasive breast imaging (57, 13.8%).

The DECT-examinations were indicated in 12 of the 415 cases due to a benign underlying disease (2.9%) and in 403 cases (97.1%) to clarify an underlying malignant disease. Staging/re-staging and therapy controls for breast carcinomas represented the largest proportion of indications (204 examinations, 49.2%), followed by staging/re-staging examinations for skin cancer (62, 15.0%), and malignant lymphoma (30, 7.2%).

The readers´ assessments are given in Figure 1 and 2.

Fig 1: Numeric frequency of the readers` ratings using a 5-point Likert scale. The results of the quantitative solely decriptor based evaluations are given on the ordinate, those of the combined approach (descripors & iodine content are summarized on the abscissa.
Fig 2: Box plot representations of the classifications of the 6 evaluators based on the descriptor evaluations (top row) and the combined qualitative & quantitative evaluations (bottom row) for benign and malignant lesions according to the reference standard.
Regarding the 763 descriptive image evaluations of benign findings according to the reference standard, there were 35 classifications (4.6%) as definitely benign, 419 classifications (54.9%) as probably benign, 24 classifications (36.6%) as equivocal, 24 classifications (3.1%) as probably malignant and 6 classifications (0.8%) as definitely malignant. This compares with 373 (48.9%) classifications as certainly benign, 200 (26.2%) classifications as probably benign, 125 (16.4%) classifications as equivocal, 55 (7.2%) classifications as probably malignant and 10 (1.3) classifications as certainly malignant for benign mass lesions according to the reference standard. The evaluation of malignant findings according to the reference standard showed a comparable results: Of the 544 image analyses confirmed as malignant, null were classified as definitely benign, 12 (2.2%) as probably benign, 105 (19.3%) as equivocal, 246 (45.2%) as probably malignant and 181 (33.3%) as definitely malignant when regarding the descriptive parameters alone. In the synoptic analysis of the descriptive and quantitative parameters, no confirmed cancer was classified as certainly benign. 14 (2.6%) confirmed malignant findings were classified as probably benign, 32 (5.9%) malignant findings as equivocal, 136 malignant findings (25.0%) as probably malignant and 362 malignant findings as certainly malignant. On 7 occasions, one study DECT-assessments of a malignant mass was changed from “equivocal” based on descriptive parameters solely in “probably benign” based on the additional information on the measured iodine content. The 7 patients in question had been referred by the University Breast Center for suspected breast malignancy and had received a breast ultrasound as initial imaging. Consistent herewith breast ultrasound was recommended as method of choice for the further diagnostic work-up in the study set-up.

Regarding the classification tree for differentiation between benign and malignant lesions the mass contour proved to be the first hierarchical decision level as visualized in Figure 3

Fig 3: DCT decision tree.
. In the second hierarchical level, automatically generated classifi­cations of the iodine content allowed to generate diagnostic predictors analogous to the BIRADS-classification. The classification tree incorporates 3 lesion descriptors with a depth of 2 ramifications (1. smooth margin; 2. irregular/unsharp margin; 3. spiculae) and iodine content adjusted to the ramifications “smooth margin” (≤0.6 mg/ml vs. >0.6 mg/ml) and irregular/unsharp margin” (≤0.2 mg/ml vs. >0.2–0.9 mg/ml vs. ≤0.9 mg/ml vs. >0.9 mg/ml) based on the calculated diagnostic specifities of each subgroup outlined in Figure 2. The inclusion of further descriptors did not result in favourable descriptor classifications.
GALLERY