Deep learning performance for detection and classification of microcalcifications on mammography.

BACKGROUND: Breast cancer screening through mammography is crucial for early detection, yet the demand for mammography services surpasses the capacity of radiologists. Artificial intelligence (AI) can assist in evaluating microcalcifications on mammography. We developed and tested an AI model for localizing and characterizing microcalcifications. METHODS: Three expert radiologists annotated a dataset of mammograms using histology-based ground truth. The dataset was partitioned for training, validation, and testing. Three neural networks (AlexNet, ResNet18, and ResNet34) were trained and evaluated using specific metrics including receiver operating characteristics area under the curve (AUC), sensitivity, and specificity. The reported metrics were computed on the test set (10% of the whole dataset). RESULTS: The dataset included 1,000 patients aged 21-73 years and 1,986 mammograms (180 density A, 220 density B, 380 density C, and 220 density D), with 389 malignant and 611 benign groups of microcalcifications. AlexNet achieved the best performance with 0.98 sensitivity, 0.89 specificity of, and 0.98 AUC for microcalcifications detection and 0.85 sensitivity, 0.89 specificity, and 0.94 AUC of for microcalcifications classification. For microcalcifications detection, ResNet18 and ResNet34 achieved 0.96 and 0.97 sensitivity, 0.91 and 0.90 specificity and 0.98 and 0.98 AUC, retrospectively. For microcalcifications classification, ResNet18 and ResNet34 exhibited 0.75 and 0.84 sensitivity, 0.85 and 0.84 specificity, and 0.88 and 0.92 AUC, respectively. CONCLUSIONS: The developed AI models accurately detect and characterize microcalcifications on mammography. RELEVANCE STATEMENT: AI-based systems have the potential to assist radiologists in interpreting microcalcifications on mammograms. The study highlights the importance of developing reliable deep learning models possibly applied to breast cancer screening. KEY POINTS: • A novel AI tool was developed and tested to aid radiologists in the interpretation of mammography by accurately detecting and characterizing microcalcifications. • Three neural networks (AlexNet, ResNet18, and ResNet34) were trained, validated, and tested using an annotated dataset of 1,000 patients and 1,986 mammograms. • The AI tool demonstrated high accuracy in detecting/localizing and characterizing microcalcifications on mammography, highlighting the potential of AI-based systems to assist radiologists in the interpretation of mammograms.

Inter-reader agreement of breast magnetic resonance imaging and contrast-enhanced mammography in breast cancer diagnosis: a multi-reader retrospective study.

OBJECTIVE: Breast magnetic resonance imaging (MRI) and contrast-enhanced mammography (CEM) are nowadays used in breast imaging but studies about their inter-reader agreement are lacking. Therefore, we compared the inter-reader agreement of CEM and MRI in breast cancer diagnosis in the same patients. METHODS: Breast MRI and CEM exams performed in a single center (09/2020-09/2021) for an IRB-approved study were retrospectively and independently evaluated by four radiologists of two different centers with different levels of experience who were blinded to the clinical and other imaging data. The reference standard was the histological diagnosis or at least 1-year negative imaging follow-up. Inter-reader agreement was examined using Cohen's and Fleiss' kappa (κ) statistics and compared with the Wald test. RESULTS: Of the 750 patients, 395 met inclusion criteria (44.5 ± 14 years old), with 752 breasts available for CEM and MRI. Overall agreement was moderate (κ = 0.60) for MRI and substantial (κ = 0.74) for CEM. For expert readers, the agreement was substantial (κ = 0.77) for MRI and almost perfect (κ = 0.82) for CEM; for non-expert readers was fair (κ = 0.39); and for MRI and moderate (κ = 0.57) for CEM. Pairwise agreement between expert readers and non-expert readers was moderate (κ = 0.50) for breast MRI and substantial (κ = 0.74) for CEM and it showed a statistically superior agreement of the expert over the non-expert readers only for MRI (p = 0.011) and not for CEM (p = 0.062). CONCLUSIONS: The agreement of CEM was superior to that of MRI (p = 0.012), including for both expert (p = 0.031) and non-expert readers (p = 0.005).

Prediction of the Malignancy of a Breast Lesion Detected on Breast Ultrasound: Radiomics Applied to Clinical Practice.

The study aimed to evaluate the performance of radiomics features and one ultrasound CAD (computer-aided diagnosis) in the prediction of the malignancy of a breast lesion detected with ultrasound and to develop a nomogram incorporating radiomic score and available information on CAD performance, conventional Breast Imaging Reporting and Data System evaluation (BI-RADS), and clinical information. Data on 365 breast lesions referred for breast US with subsequent histologic analysis between January 2020 and March 2022 were retrospectively collected. Patients were randomly divided into a training group (n = 255) and a validation test group (n = 110). A radiomics score was generated from the US image. The CAD was performed in a subgroup of 209 cases. The radiomics score included seven radiomics features selected with the LASSO logistic regression model. The multivariable logistic model incorporating CAD performance, BI-RADS evaluation, clinical information, and radiomic score as covariates showed promising results in the prediction of the malignancy of breast lesions: Area under the receiver operating characteristic curve, [AUC]: 0.914; 95% Confidence Interval, [CI]: 0.876-0.951. A nomogram was developed based on these results for possible future applications in clinical practice.

Women's perceptions and attitudes to the use of AI in breast cancer screening: a survey in a cancer referral centre.

OBJECTIVE: Although breast cancer screening can benefit from Artificial Intelligence (AI), it is still unknown whether, to which extent or under which conditions, the use of AI is going to be accepted by the general population. The aim of our study is to evaluate what the females who are eligible for breast cancer screening know about AI and how they perceive such innovation. METHODS: We used a prospective survey consisting of a 11-multiple-choice questionnaire evaluating statistical associations with Chi-Square-test or Fisher-exact-test. Multinomial-logistic-regression was performed on items with more than two response categories. Odds ratio (OR) with 95% CI were computed to estimate the probability of a specific response according to patient's characteristics. RESULTS: In the 800 analysed questionnaires, 51% of respondents confirmed to have knowledge of AI. Of these, 88% expressed a positive opinion about its use in medicine. Non-Italian respondents were associated with the belief of having a deep awareness about AI more often than Italian respondents (OR = 1.91;95% CI[1.10-3.33]). Higher education level was associated with better opinions on the use of AI in medicine (OR = 4.69;95% CI[1.36-16.12]). According to 94% of respondents, the radiologists should always produce their own report on mammograms, whilst 77% agreed that AI should be used as a second reader. Most respondents (52%) considered that both the software developer and the radiologist should be held accountable for AI errors. CONCLUSIONS: Most of the females undergoing screening in our Institute approve the introduction of AI, although only as a support to radiologist, and not in substitution thereof. Yet, accountability in case of AI errors is still unsolved. advances in knowledge:This survey may be considered as a pilot-study for the development of large-scale studies to understand females's demands and concerns about AI applications in breast cancer screening.

Contrast-Enhanced Spectral Mammography and tumor size assessment: a valuable tool for appropriate surgical management of breast lesions.

PURPOSE: To compare the accuracy of Contrast-Enhanced Spectral Mammography (CESM), MG, US, and breast MRI in estimating the size of breast lesions requiring surgery. The postoperative histology size of the lesion was used as the gold standard. MATERIAL AND METHODS: Two hundred thirty-three non-benign lesions in 189 patients were included in the analyses. All the selected patients underwent CESM and at least one other conventional diagnostic exam (US, MG, or MRI). Subsequently, all the patients underwent surgery preceded by cytological/histological examination. The largest diameter of the lesion at imaging was measured by a radiologist with more than 10 years' experience and then compared with the size of the lesion in the histological sample at the surgery (gold standard). RESULTS: Among the 233 breast lesions, 196 were evaluated with US, 206 with MG and 160 with MRI. We found no statistically significant differences between size measurements using CESM and MRI compared with the measurements at the surgery (p value 0.63 and 0.51), whereas a significant difference was found for MG and US (p < 0.001). CONCLUSION: CESM is a reliable method for estimating the size of breast lesions: its performance seems superior to US and MG and comparable to MRI.

A Score to Predict the Malignancy of a Breast Lesion Based on Different Contrast Enhancement Patterns in Contrast-Enhanced Spectral Mammography.

Background: To create a predictive score of malignancy of a breast lesion based on the main contrast enhancement features ascertained by contrast-enhanced spectral mammography (CESM). Methods: In this single-centre prospective study, patients with suspicious breast lesions (BIRADS > 3) were enrolled between January 2013 and February 2022. All participants underwent CESM prior to breast biopsy, and eventually surgery. A radiologist with 20 years' experience in breast imaging evaluated the presence or absence of enhancement and the following enhancement descriptors: intensity, pattern, margin, and ground glass. A score of 0 or 1 was given for each descriptor, depending on whether the enhancement characteristic was predictive of benignity or malignancy (both in situ and invasive). Then, an overall enhancement score ranging from 0 to 4 was obtained. The histological results were considered the gold standard in the evaluation of the relationship between enhancement patterns and malignancy. Results: A total of 321 women (median age: 51 years; range: 22−83) with 377 suspicious breast lesions were evaluated. Two hundred forty-nine lesions (66%) have malignant histological results (217 invasive and 32 in situ). Considering an overall enhancement score ≥ 2 as predictive of malignancy, we obtain an overall sensitivity of 92.4%; specificity of 89.8%; positive predictive value of 94.7%; and negative predictive value of 85.8%. Conclusions: Our proposed predictive score on the enhancement descriptors of CESM to predict the malignancy of a breast lesion shows excellent results and can help in early breast cancer diagnosis and in avoiding unnecessary biopsies.

Evaluation of computer-aided diagnosis in breast ultrasonography: Improvement in diagnostic performance of inexperienced radiologists.

PURPOSE: To evaluate if a computer-aided diagnosis (CAD) system on ultrasound (US) can improve the diagnostic performance of inexperienced radiologists. METHODS: We collected ultrasound images of 256 breast lesions taken between March and May 2020. We asked two experienced and two inexperienced radiologists to retrospectively review the US features of each breast lesion according to the Breast Imaging Reporting and Data System (BI-RADS) categories. A CAD examination with S-Detect™ software (Samsung Healthcare, Seoul, South Korea) was conducted retrospectively by another uninvolved radiologist blinded to the BIRADS values previously attributed to the lesions. Diagnostic performances of experienced and inexperienced radiologists and CAD were compared and the inter-observer agreement among radiologists was calculated. RESULTS: The diagnostic performance of the experienced group in terms of sensitivity was significantly higher than CAD (p < 0.001). Conversely, the diagnostic performance of inexperienced group in terms of both sensitivity and specificity was significantly lower than CAD (p < 0.001). We obtained an excellent agreement in the evaluation of the lesions among the two expert radiologists (Kappa coefficient: 88.7%), and among the two non-expert radiologists (Kappa coefficient: 84.9%). CONCLUSION: The US CAD system is a useful additional tool to improve the diagnostic performance of the inexperienced radiologists, eventually reducing the number of unnecessary biopsies. Moreover, it is a valid second opinion in case of experienced radiologists.

Atypical Ductal Hyperplasia after Vacuum-Assisted Breast Biopsy: Can We Reduce the Upgrade to Breast Cancer to an Acceptable Rate?

(1) Background: to evaluate which factors can reduce the upgrade rate of atypical ductal hyperplasia (ADH) to in situ or invasive carcinoma in patients who underwent vacuum-assisted breast biopsy (VABB) and subsequent surgical excision. (2) Methods: 2955 VABBs were reviewed; 141 patients with a diagnosis of ADH were selected for subsequent surgical excision. The association between patients' characteristics and the upgrade rate to breast cancer was evaluated in both univariate and multivariate analyses. (3) Results: the upgrade rates to ductal carcinoma in situ (DCIS) and invasive carcinoma (IC) were, respectively, 29.1% and 7.8%. The pooled upgrade rate to DCIS or IC was statistically lower at univariate analysis, considering the following parameters: complete removal of the lesion (p-value < 0.001); BIRADS ≤ 4a (p-value < 0.001); size of the lesion ≤15 mm (p-value: 0.002); age of the patients <50 years (p-value: 0.035). (4) Conclusions: the overall upgrade rate of ADH to DCIS or IC is high and, as already known, surgery should be recommended. However, ADH cases should always be discussed in multidisciplinary meetings: some parameters appear to be related to a lower upgrade rate. Patients presenting these parameters could be strictly followed up to avoid overtreatment.

Inter-Reader Agreement of Diffusion-Weighted Magnetic Resonance Imaging for Breast Cancer Detection: A Multi-Reader Retrospective Study.

PURPOSE: In order to evaluate the use of un-enhanced magnetic resonance imaging (MRI) for detecting breast cancer, we evaluated the accuracy and the agreement of diffusion-weighted imaging (DWI) through the inter-reader reproducibility between expert and non-expert readers. MATERIAL AND METHODS: Consecutive breast MRI performed in a single centre were retrospectively evaluated by four radiologists with different levels of experience. The per-breast standard of reference was the histological diagnosis from needle biopsy or surgical excision, or at least one-year negative follow-up on imaging. The agreement across readers (by inter-reader reproducibility) was examined for each breast examined using Cohen's and Fleiss' kappa (κ) statistics. The Wald test was used to test the difference in inter-reader agreement between expert and non-expert readers. RESULTS: Of 1131 examinations, according to our inclusion and exclusion criteria, 382 women were included (49.5 ± 12 years old), 40 of them with unilateral mastectomy, totaling 724 breasts. Overall inter-reader reproducibility was substantial (κ = 0.74) for expert readers and poor (κ = 0.37) for non- expert readers. Pairwise agreement between expert readers and non-expert readers was moderate (κ = 0.60) and showed a statistically superior agreement of the expert readers over the non-expert readers (p = 0.003). CONCLUSIONS: DWI showed substantial inter-reader reproducibility among expert-level readers. Pairwise comparison showed superior agreement of the expert readers over the non-expert readers, with the expert readers having higher inter-reader reproducibility than the non-expert readers. These findings open new perspectives for prospective studies investigating the actual role of DWI as a stand-alone method for un-enhanced breast MRI.

COVID-19 and breast fine needle aspiration cytology method: What should we change?

INTRODUCTION: Air-dried slide preparation for fine needle aspiration cytology procedures is currently considered unsafe because of the risk of infectious aerosols of coronavirus 19. This study compares the safety and accuracy of two different protocols, one with and one without air-dried slides. METHODS: Starting from 3 March 2020, we discontinued the use of air-dried slides during breast fine needle aspiration procedures. We selected cases collected during two periods: 2 months before and 2 months after 3 March. In both groups, the number of procedures was recorded together with the distribution of the diagnostic categories and the concordance between cytological and histological results on surgical specimens for lesions suggestive of malignancy, using the chi-squared test. RESULTS: Of the 100 procedures performed during the pre-COVID-19 period, 55% were negative (C2), 3% were non-diagnostic (C1) and 40% were positive (C4 or C5). Of the 75 procedures obtained during the COVID-19 period, 44% were negative (C2), 2.7% were non-diagnostic (C1) and 52% were positive (C4 or C5). Despite the use of a new protocol during the COVID-19 period, we observed concordance between cytological and histological results for lesions suggestive of malignancy. There was no statistically significant difference concerning the distribution of the diagnostic categories in the two groups. CONCLUSIONS: Taking into account the slightly lower number of procedures being analysed during the COVID-19 period, the introduction of a new protocol that does not include air-dried slides is safe and reliable.

Myeloid sarcoma of the breast: a pathology that should not be forgotten.

Myeloid sarcoma (MS) is a rare neoplasm, represented by a tumoural mass composed of myeloid blasts, occurring at any anatomical site other than the bone marrow. MS is considered the tissue-based equivalent of acute myeloid leukaemia (AML), requiring the same therapeutic specification, independently from the association with previous or coexisting myeloid neoplasms. Isolated breast involvement by MS is exceedingly rare, with only exceptional cases reported in the literature. This work aims to provide a pictorial essay of the main features of breast involvement by MS. Even though it is a rare condition, we should not forget this neoplasm, and its possibility of being disguised by the AML, as it requires prompt treatment.