Search

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has a high sensitivity in detecting breast cancer but often leads to unnecessary biopsies and patient workup. We used a deep learning (DL) system to improve the overall accuracy of breast cancer diagnosis and personalize management of patients undergoing DCE-MRI. On the internal test set ( n = 3936 exams), our system achieved an area under the receiver operating characteristic curve (AUROC) of 0.92 (95% CI: 0.92 to 0.93). In a retrospective reader study, there was no statistically significant difference ( P = 0.19) between five board-certified breast radiologists and the DL system (mean ΔAUROC, +0.04 in favor of the DL system). Radiologists’ performance improved when their predictions were averaged with DL’s predictions [mean ΔAUPRC (area under the precision-recall curve), +0.07]. We demonstrated the generalizability of the DL system using multiple datasets from Poland and the United States. An additional reader study on a Polish dataset showed that the DL system was as robust to distribution shift as radiologists. In subgroup analysis, we observed consistent results across different cancer subtypes and patient demographics. Using decision curve analysis, we showed that the DL system can reduce unnecessary biopsies in the range of clinically relevant risk thresholds. This would lead to avoiding biopsies yielding benign results in up to 20% of all patients with BI-RADS category 4 lesions. Last, we performed an error analysis, investigating situations where DL predictions were mostly incorrect. This exploratory work creates a foundation for deployment and prospective analysis of DL-based models for breast MRI.

Purpose Breast cancer is the most common cause of death from neoplastic disease in women. Among all breast anatomy types, glandular type is the most problematic concerning evaluation. While digital mammography still remains the basic diagnostic tool, one must be aware of its limitations in dense breasts. Although magnetic resonance imaging (MRI) has greatly improved sensitivity, its specificity is low. Moreover, there are contraindications for MRI for some patients, so a substitute has been searched for. This study was performed to check if contrast-enhanced spectral mammography (CESM) can be a viable option for patients with dense breasts. Material and methods The study involved 121 patients with abnormalities detected on base-line diagnostic imaging (ultrasound or mammography). The patients had subsequent examinations, both CESM and MRI performed within a maximum 2-month time interval. The sensitivity and specificity of both methods in the whole group as well as in specific breast structure types were measured and compared. Results Contrast enhancement was visible in all 121 cases on MRI, while on CESM lack of enhancement was noted in 13 cases. All of those 13 lesions turned out to be benign. There were 40 (33%) benign and 81 (69%) malignant tumours. The analysed group included 53 (44%) glandular type breast patients, 39 (32%) mixed type, and 29 (23%) fatty type. Although MRI proved to be slightly more effective in dense breasts, both methods showed similar results in the whole study group. Conclusion CESM can be used with confidence in patients with glandular breast type when MRI is not available or there are reported contraindications to MRI.

Background/aim Contrast enhanced spectral mammography (CESM) is a novel method of breast cancer diagnosis. Benign lesions are enhanced after contrast injection on both CESM and breast magnetic resonance imaging (MRI). Kinetic curves on breast MRI facilitate differentiation between benign and malignant lesions, while on CESM there is no such possibility and we need to asses lesions based only on their level of enhancement and its patterns. The aim of this study was to compare two subjective assessments of enhancement level on CESM with numerical values measured using the region of interests (ROIs) and to categorize obtained enhancement level values. Patients and methods Patients with suspicious findings on previously performed examinations were qualified for CESM. The qualitative assessment was based on the subjective classification of visible contrast enhancement as weak, medium or strong. The quantitative assessment was obtained by measurements of an average enhancement value and sigma value within the ellipsoidal shape ROI, inserted into the evaluated contrast enhancing lesion. Results The study group included 151 patients with total of 195 lesions diagnosed. It was verified how the classification based on the threshold values of %RS (percentage signal difference between enhancing lesion and background) and SDNR (signal-difference-to-noise Ratio) corresponds to a subjective assessment. Conclusion Quantitative assessment of contrast enhancement on CESM is helpful in making decisions whether a lesion requires a biopsy. This can reduce the number of unnecessary biopsy procedures and reduce the cost of diagnostics.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has a very high sensitivity in detecting breast cancer, but it often leads to unnecessary biopsies and patient workup. In this paper, we used an artificial intelligence (AI) system to improve the overall accuracy of breast cancer diagnosis and personalize management of patients undergoing DCE-MRI. On the internal test set (N=3,936 exams), our system achieved an area under the receiver operating characteristic curve (AUROC) of 0.92 (95% CI: 0.92-0.93). In a retrospective reader study, there was no statistically significant difference between 5 board-certified breast radiologists and the AI system (mean ΔAUROC +0.04 in favor of the AI system). Radiologists’ performance improved when their predictions were averaged with AI’s predictions (mean ΔAUPRC [area under the precision-recall curve] +0.07). Those hybrid predictions also increase interreader agreement (Fleiss’ kappa Δ +0.21 (0.16-0.26)). We demonstrated the generalizability of the AI system using multiple data sets from Poland and the US. In subgroup analysis, we observed consistent results across different cancer subtypes and patient demographics. Using the decision curve analysis, we showed that the AI system can reduce unnecessary biopsies in the range of clinically relevant risk thresholds. This would lead to avoiding benign biopsies in up to 20% of all BI-RADS category 4 patients. Finally, we performed an error analysis, investigating situations where AI predictions were mostly incorrect. This exploratory work creates a foundation for deployment and prospective analysis of AI-based models for breast MRI.One Sentence SummaryWe developed and evaluated an AI system for predicting breast cancer in MRI that matches the performance of board-certified radiologists and has the potential to prevent unnecessary biopsies.

BACKGROUND This retrospective study included 103 patients diagnosed with rectal adenocarcinoma at a single center in Poland who underwent preoperative diffusion-weighted magnetic resonance imaging (DWI) and aimed to determine whether the apparent diffusion coefficient (ADC) was an imaging marker for tumor invasion and regional lymph node involvement. MATERIAL AND METHODS We analyzed primary staging magnetic resonance examinations of the rectum of 103 consecutive patients with histologically proven non-mucinous adenocarcinoma who underwent surgical treatment. In 85 patients, surgery was preceded by long-course chemoradiotherapy (n=18) or short-course radiotherapy (n=67). The following DWI parameters were measured: ADC mean, minimum, maximum, and standard deviation in the region of interest (ADC SD-in-ROI). Values were compared between subgroups based on histological parameters from the report: tumor stage, lymph node stage, differentiation grade, the presence of extranodal tumor deposits, angioinvasion, and perineural invasion. Statistical analysis was performed using the Mann-Whitney U test and the unilateral t test. RESULTS ADC mean values were lower for cases in which postoperative histopathological examination lymph node invasion (P=0.04) and tumor deposits were found (P=0.04). Minimal ADC value was higher in cases in which tumor deposits were not found (P=0.009). ADC SD-in-ROI values were lower in cases in which lymph nodes invasion was confirmed (P=0.014). There were no statistically significant differences for other parameters. CONCLUSIONS The ADC values in pre-treatment DWI in patients with rectal adenocarcinoma were correlated with tumor invasion and regional lymph node metastases. Therefore, ADC values from the pre-treatment MRI may help plan adjuvant therapy in patients with rectal adenocarcinoma.

Breast cancer, which is the most common cancer in women, is a major problem both in Poland and worldwide. Mammography remains the primary screening method. However, the sensitivity of mammographic screening is lower in women with dense glandular breasts due to tissue overlap and the effect of the glandular tissue obscuring the tumor and the fact that tumors and glandular tissue show similar X-ray absorption. Consequently, other methods are being sought to increase breast cancer detection rates. Currently, the most common and used methods are ultrasonography, magnetic resonance imaging and advanced mammographic methods (digital breast tomosynthesis and contrast-enhanced spectral mammography). Despite many advantages and superiority over mammography in dense breasts, they also have many disadvantages. Ultrasound is operator-dependent and the other techniques are expensive or not widely available. The Automated Breast Ultrasound Service (ABUS) technique appears to be a good option in terms of both effectiveness and lower cost.

Contrast-enhanced spectral mammography (CESM) is a novel method for breast cancer detection. The aim of this study is to check if there is a possibility of quantitative assessment of contrast enhancement in CESM and if there is any correlation between quantitative assessment of contrast enhancement in CESM and histopathology. A total of 167 female patients underwent CESM. All subjects previously had suspicious lesions found on mammography, breast ultrasound, or both. After imaging, the following parameters were evaluated: number of enhancing lesions in each breast and size and degree of enhancement of each lesion. Based on the collected data, the percentage signal difference between enhancing lesion and background (%RS) and signal-difference-to-noise ratio (SDNR) were measured for each lesion. The number of lesions detected in the study population was 195. Among all diagnosed lesions, 120 (62%) were assessed to be infiltrating cancers, 16 (8%) non-infiltrating cancers, and 59 (30%) were benign. Thirteen (7%) lesions did not enhance in CESM; all non-enhancing lesions were confirmed to be benign under histopathological examination. Analysis of enhancement indices showed that signal values within lesions and signal values within background ROIs (regions of interest) were similar in CC (craniocaudal) and MLO (mediolateral) projections. Mean %RS values were correlated with the type of enhancing lesion, infiltrating cancers having the highest values, benign lesions the lowest. This work has demonstrated a significant correlation between the degree of lesion enhancement in CESM and malignancy. Quantitative analysis of enhancement levels in CESM can distinguish between invasive cancers and benign or in situ lesions. • There is a possibility of quantitative assessment of contrast enhancement in CESM. • Correlation between quantitative assessment of contrast enhancement in CESM and histopathology was observed.

Objectives: With the growing number of new breast cancer cases in women, new methods of imaging arise. Contrast enhanced spectral mammography (CESM) and magnetic resonance imaging (MRI) are comparable methods regarding sensitivity. The aim of this study is to check if analysis of background parenchymal enhancement on CESM can improve its usefulness. Material and methods: A total of 64 patients with breast lesions found previously on ultrasound or mammography underwent MRI and CESM within less than one month. On MRI the contrast enhancement kinetics and visual BPE were evaluated. On CESM the enhancement of lesions was noted as well as a quantitative level of BPH. The gathered data was analysed in terms of patterns and relations. Results: A total of 66 lesions were identified both on MRI and CESM, including 11 (17%) benign and 55 (83%) malignant lesions. Among malignant lesions 13 (20%) were assessed as intraductal and 42 (64%) as infiltrating carcinomas. The study showed correlation between the level of enhancement on CESM and the type of kinetic curve on MRI and lesion enhance-ment on CESM as well as confirmed the fact that the BPE is a destimulant in both methods of imaging. Conclusions: Evaluation of BPE level on CESM can help reading radiologists to define a lesion as malignant with higher probability than based only on the qualitative lesion enhancement level.

A case report of bilateral primary angiosarcoma treated with neoadjuvant chemotherapy was presented. A routine diagnostic mammography and ultrasound examinations indicated abnormalities in both breasts of the patient, confirmed on MRI as large bilateral masses. Core needle biopsy revealed angiosarcoma G1. The treatment agreed during the interdisciplinary meeting involved chemotherapy combined with simultaneous blockade of beta-adrenergic receptors, followed by bilateral simple mastectomy. This case highlights the importance of a patient-focused care.

strongBACKGROUND/strongDue to the decreased sensitivity of mammography in glandular breasts, new diagnostic modalities, like contrast-enhanced spectral mammography (CESM) and digital breast tomosynthesis (DBT) have been developed. The aim of this study was to compare qualitative enhancement levels on CESM with type of kinetic enhancement curves on MRI examination.strongMATERIAL AND METHODS/strongPatients qualified for the CESM examination presented some diagnostic doubts - suspected multifocality, multicentricity, or having dense glandular breast tissue. The enhancement level on CESM was described as weak, medium, or strong. Enhancement on achieved MR images was assessed on the basis of enhancement kinetic curves. The level of enhancement on CESM was associated with enhancement curves type on MRI. All lesions detected on CESM and MRI were histopathologically verified.strongRESULTS/strongThe study involved 107 lesions diagnosed in 94 patients: 71 lesions (66%) appeared to be infiltrating on histopathological examination, 9 lesions (8%) were non-infiltrating cancers, and 27 lesions (25%) were benign. Data analysis revealed that lesions with wash-out curve on MRI most often presented strong enhancement on CESM, while in lesions with progressive enhancement curve, strong enhancement on CESM was the rarest. The relationship between enhancement level on CESM and curve type on contrast-enhanced MRI depends on the nature of the lesion. The type of MRI curve was found to be associated with enhancement level on CESM.strongCONCLUSIONS/strongWe compared subjective assessments of contrast enhancement on CESM with enhancement kinetic curves on MRI. The results showed that the level of enhancement on CESM and type of kinetic curves on MRI depends on the lesion type.

BACKGROUND/AIM It is possible that the degree of enhancement on contrast-enhanced spectral mammography (CESM), a new diagnostic method, might provide prognostic information for breast cancer patients. Therefore, in a group of 82 breast cancer patients, we analyzed the prognostic significance of degree and pattern of enhancement on CESM as well as its relation to: (a) breast cancer immunophenotype (based on ER/PR/HER2 status) (b) podoplanin expression in cancer stroma (lymphatic vessel density plus podoplanin-positivity of cancer-associated fibroblasts), and (c) other histological parameters. MATERIALS AND METHODS For each tumor the intensity of enhancement on CESM was qualitatively assessed as strong or weak/medium, while the pattern - as homogenous and heterogenous. RESULTS Herein we report, for the first time, that strong and heterogenous enhancement on CESM was related to unfavorable disease-free survival of breast cancer patients (p=0.005). Moreover, the strong enhancement was more frequent in large and node-positive tumors (pT>1, pN>0) (p=0.002), as well as in carcinomas with podoplanin-sparse stroma (p=0.008). CONCLUSION Intensity and pattern of enhancement on CESM might provide (together with the results of other diagnostic imaging methods) not only the confirmation of presence or absence of tumor, but also prognostic information.