ABSTRACT
OBJECTIVES: One of the most promising adjuncts for screening breast cancer is ultrasound (US) radio-frequency (RF) time series. It has the superiority of not requiring any supplementary equipment over other methods. This research aimed to propound a machine learning (ML) approach for automatically classifying benign, probably benign, suspicious, and malignant breast lesions based on the features extracted from the accumulated US RF time series. METHODS: In this article, 220 data of the aforementioned categories, recorded from 118 patients, were analyzed. The dataset, named RFTSBU, was registered by a SuperSonic Imagine Aixplorer medical/research system equipped with a linear transducer. The regions of interest (ROIs) of the B-mode images were manually selected by an expert radiologist before computing the suggested features. Regarding time, frequency, and time-frequency domains, 291 various features were extracted from each ROI. Finally, the features were classified by a pioneering technique named the reference classification method (RCM). Furthermore, the Lee filter was applied to evaluate the effectiveness of reducing speckle noise on the outcomes. RESULTS: The accuracy of two-class, three-class, and four-class classifications were respectively calculated 98.59 ± 0.71%, 98.13 ± 0.69%, and 96.10 ± 0.66% (considering 10 repetitions) while support vector machine (SVM) and K-nearest neighbor (KNN) classifiers with 5-fold cross-validation were utilized. CONCLUSIONS: This article represented the proposed approach, named CCRFML, to distinguish between breast lesions based on registered in vivo RF time series employing an ML framework. The proposed method's impressive level of classification accuracy attests to its capability of effectively assisting medical professionals in the noninvasive differentiation of breast lesions.
ABSTRACT
PURPOSE: This study aims to compare the accuracy of the ADNEX MR scoring system and pattern recognition system to evaluate adnexal lesions indeterminate on the US exam. METHODS: In this cross-sectional retrospective study, pelvic DCE-MRI of 245 patients with 340 adnexal masses was studied based on the ADNEX MR scoring system and pattern recognition system. RESULTS: ADNEX MR scoring system with a sensitivity of 96.6% and specificity of 91% has an accuracy of 92.9%. The pattern recognition system's sensitivity, specificity, and accuracy are 95.8%, 93.3%, and 94.7%, respectively. PPV and NPV for the ADNEX MR scoring system were 85.1 and 98.1, respectively. PPV and NPV for the pattern recognition system were 89.7% and 97.7%, respectively. The area under the ROC curve for the ADNEX MR scoring system and pattern recognition system is 0.938 (95% CI, 0.909-0.967) and 0.950 (95% CI, 0.922-0.977). Pairwise comparison of these AUCs showed no significant difference (p = 0.052). CONCLUSION: The pattern recognition system is less sensitive than the ADNEX MR scoring system, yet more specific.
Subject(s)
Adnexal Diseases , Magnetic Resonance Imaging , Humans , Female , Cross-Sectional Studies , Retrospective Studies , Middle Aged , Adnexal Diseases/diagnostic imaging , Adnexal Diseases/pathology , Adnexal Diseases/diagnosis , Adult , Magnetic Resonance Imaging/methods , Aged , Prognosis , ROC Curve , Follow-Up Studies , Adolescent , Young Adult , Pattern Recognition, Automated/methods , Adnexa Uteri/pathology , Adnexa Uteri/diagnostic imagingABSTRACT
Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is a highly sensitive breast imaging modality in detecting breast carcinoma. Nonmass enhancement (NME) is uniquely seen on MRI of the breast. The correlation between NME features and pathologic results has not been extensively explored. Our goal was to evaluate the characteristics of probably benign and suspicious NME lesions in MRI and determine which features are more associated with malignancy. We performed a retrospective research after approval by the hospital ethics committee on women who underwent breast MRI from March 2017 to March 2020 and identified 63 lesions of all 400 NME that were categorized as probably benign or suspicious according to the BI-RADS classification (version 2013). MRI features of NME findings including the location, size, distribution and enhancement pattern, kinetic curve, diffusion restriction, and also pathology result or 6-12-month follow-up MRI were evaluated and analyzed in each group (probably benign or suspicious NME). Vacuum-guided biopsies (VAB) were performed under mammographic or sonographic guidance and confirmed with MRI by visualization of the inserted clips. Segmental distribution and clustered ring internal enhancement were significantly associated with malignancy (p value<0.05), while linear distribution or homogeneous enhancement patterns were associated with benignity (p value <0.05). Additionally, the plateau and washout types in the dynamic curve were only seen in malignant lesions (p value <0.05). The presence of DWI restriction in NME lesions was also found to be a statistically important factor. Understanding the imaging findings of malignant NME is helpful to determine when biopsy is indicated. The correlation between NME features and pathologic results is critical in making appropriate management.
ABSTRACT
Background and Aims: Data mining methods are effective and well-known tools for developing predictive models and extracting useful information from various data of patients. The present study aimed to predict the severity of patients with COVID-19 by applying the rule mining method using characteristics of medical images. Methods: This retrospective study has analyzed the radiological data from 104 COVID-19 hospitalized patients diagnosed with COVID-19 in a hospital in Iran. A data set containing 75 binary features was generated. Apriori method is utilized for association rule mining on this data set. Only rules with confidence equal to one were generated. The performance of rules is calculated by support, coverage, and lift indexes. Results: Ten rules were extracted with only X-ray-related features on cases referred to ICU. The Support and Coverage index of all of these rules was 0.087, and the Lift index of them was 1.58. Thirteen rules were extracted from only CT scan-related features on cases referred to ICU. The CXR_Pleural effusion feature has appeared in all the rules. The CXR_Left upper zone feature appears in 9 rules out of 10. The Support and Coverage index of all rules was 0.15, and the Lift index of all rules was 1.63. the CT_Adjacent pleura thickening feature has appeared in all rules, and the CT_Right middle lobe appeared in 9 rules out of 13. Conclusion: This study could reveal the application and efficacy of CXR and CT scan imaging modalities in predicting ICU admission to a major COVID-19 infection via data mining methods. The findings of this study could help data scientists, radiologists, and clinicians in the future development and implementation of these methods in similar conditions and timely and appropriately save patients from adverse disease outcomes.
ABSTRACT
Background: Although, there are accumulating evidence about diagnostic role of abbreviated breast magnetic resonance imaging (MRI) in screening setting, the implementation of abbreviated MRI in staging of breast cancer has been poorly elucidated. Objective: To evaluate the diagnostic performance of abbreviated breast MRI in estimating extent of disease before initiation of neoadjuvant chemotherapy. Methods: A total of 54 patients with biopsy-proven main lesion referred to evaluate by standard protocol breast MRI before initiation of neoadjuvant chemotherapy were retrospectively enrolled. From a standard protocol, a data set of abbreviated protocol consisting fat-saturated T1-weighted (T1W) pre-contrast and first two fat-saturated T1W post-contrast series with reconstruction of their subtraction including maximum intensity projection (MIP) were obtained and interpreted. The concordance rate of abbreviated with standard protocol (as a reference standard) were compared. Diagnostic accuracy, sensitivity, specificity, and positive and negative predictive value were calculated, as well. Results: The maximum size of the main mass was 38.6 ± 17.3 and 40.7 ± 17.9 for abbreviated and standard protocol, respectively. All of the main mass was detected by abbreviated protocol with 100% concordance. Concordance was 98.1% and 94.4% in terms of multifocal/multicentric status and for estimating of NME, respectively. The abbreviated protocol has high sensitivity and specificity with more than 90% value regarding main mass detection, measurement of the maximum size of the main mass, determination of multifocal/multicenter status and NAC involvement. Conclusion: Abbreviated protocol may be a reliable surrogate for standard protocol breast MRI in evaluating extent of breast cancer.
ABSTRACT
BACKGROUND: Compared to other breast cancer subtypes, triple-negative breast cancer (TNBC) has always been challenging for clinicians due to its aggressive behavior and lack of a specific treatment. There is a confirmed association between invasive features of tumors and increased epithelial-mesenchymal transition (EMT) process, which is consistent with a higher rate of EMT in TNBC. METHODS AND RESULTS: We investigated the expression of EMT-related genes, SNAI1 and MMP7, and EMT-related lncRNAs, treRNA and SBF2-AS1, in 50 TNBC tumors and 50 non-TNBC tumors to reveal more regulators and effectors involved in TNBC malignancy. In the present study, we showed the overexpression of all the studied genes and lncRNAs in TNBC tumors compared to non-TNBC samples. Moreover, a significant association was observed between MMP7 and treRNA expression levels and larger tumor size. A positive correlation between SNAI1 and lncRNA treRNA expression levels was also detected. CONCLUSIONS: Due to the differential expression and the potential diagnostic power of the studied genes, SBF2-AS1 and treRNA can be proposed as new probable biomarkers and therapeutic targets in TNBC.
Subject(s)
MicroRNAs , RNA, Long Noncoding , Triple Negative Breast Neoplasms , Humans , RNA, Long Noncoding/genetics , RNA, Long Noncoding/metabolism , Matrix Metalloproteinase 7/genetics , Triple Negative Breast Neoplasms/pathology , Gene Expression Regulation, Neoplastic/genetics , Cell Line, Tumor , Cell Proliferation/genetics , MicroRNAs/genetics , Epithelial-Mesenchymal Transition/geneticsABSTRACT
BACKGROUND: Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes coronavirus disease 2019 (COVID-19) with significant morbidity and mortality. We reported and compared the clinical and para-clinical findings of immunocompromised and immunocompetent COVID-19 patients in a case-control study at the Imam Khomeini hospital in Tehran, Iran. METHODS: In this study, 107 immunocompromised COVID-19 patients were recruited as the case group, and 107 immunocompetent COVID-19 patients as the control group. The participants were matched based on age and sex. The patients' information was retrieved from the hospital records in an information sheet. Associations between clinical and para-clinical findings with the immune status were assessed using bivariate and multivariate analyses. RESULTS: The initial pulse rate and recovery time were significantly higher in immunocompromised patients (p < .05). Myalgia, nausea/vomiting, loss of appetite, headache, and dizziness were more frequently reported by the control group (p < .05). Regarding the prescribed medications' duration, Sofosbovir was used longer in the case group, while Ribavirin was used longer in the control groups (p < .05). The most common complication in the case group was acute respiratory distress syndrome, although no major complications were observed in the control group. According to the multivariate analysis, recovery time and Lopinavir/Ritonavir (Kaletra) prescription were significantly higher in the immunocompromised compared to the immunocompetent group. CONCLUSION: Recovery time was significantly longer in the immunocompromised compared to the immunocompetent group, which emphasizes the necessity of prolonged care in these high-risk patients. Also, it is recommended to investigate the effect of novel therapeutic interventions to reduce the recovery time in addition to improving the prognosis of immunodeficient patients with COVID-19.
Subject(s)
COVID-19 , Humans , Antiviral Agents/therapeutic use , SARS-CoV-2 , Case-Control Studies , Iran/epidemiology , Immunocompromised HostABSTRACT
Background: Synthesized Mammogram (SM) from Digital Breast Tomosynthesis (DBT) images is introduced to replace the routine Full Field Digital Mammography (FFDM) to reduce radiation dose. Purpose: to compare the conspicuity of cancer related findings between SM and FFDM and combination of these methods with DBT. Methods: The study was conducted in a tertiary breast imaging center, where 200 women referred for screening were enrolled in the study sequentially. Patients underwent FFDM and DBT simultaneously and a two-year follow-up was done. Data was evaluated for Breast Imaging Reporting and Data System (BI-RADS) score, breast density, mass lesions, calcification, and focal asymmetry by two expert breast radiologists. Comparison between different methods was made by Cohen Kappa test. Results: 22 patients with likely malignant findings went under biopsy. Taking histopathologic findings and two-year follow up as reference, the overall sensitivity and specificity for FFDM+DBT (86.1 and 88.9 respectively) and SM+DBT (86.1 and 88.2) didn't show a meaningful difference. Comparing SM and FFDM, calcification in 20 subjects were overlooked on SM, but later detected when combined with DBT. Considering breast composition and BI-RADS categorization, an excellent agreement existed between the readers. Conclusion: Screening with SM+DBT shows comparable results with FFDM+DBT considering BI-RADS categorization of the patients. Although SM showed slightly inferior sensitivity compared to FFDM, after combining DBT with SM no malignant appearing calcification or mass lesion was missed.
ABSTRACT
Background: Ultrasound-detected breast lesions with probably benign features are a great challenge for clinicians, especially in breasts with dense composition. We aimed to investigate the finding of two radiologic modalities on these lesions. Methods: This retrospective cross-sectional study recruited patients including (1) candidates of assisted reproductive therapy (ART), (2) patients with prior high-risk lesions, and (3) the "suspected" BIRADS-3 masses referring to masses that US BIRADS-3 was not compatible with the clinical breast exam. The degree of agreement in diagnosing BIRADS-3 lesions between two modalities of magnetic resonance imaging (MRI) and ultrasonography (US), and comparison of the lesions in US and MRI were the study variables. Results: A total number of 123 lesions in 67 patients with a median age of 38 (IQR: 11, range: 17-67). In the examination by MRI, 107 (87.0 %) lesions were BIRADS-3 indicating the agreement level between these two modalities. The median size of the lesions in US was 9 mm (IQR: 5, range: 3-43) and 9 mm (IQR: 10, range: 4-46) in MRI. The measured size of the lesions between the two modalities was highly correlated (Spearman correlation coefficient: 0.889, P-value < 0.001). MRI evaluation revealed two cases of deep lesions which were missed in the US imaging. Conclusions: This study found relatively high agreement values between US and MRI in detecting BIRADS-3 breast lesions in candidates for ART or patients with prior high-risk lesions. Also, MRI could downgrade about one-tenth of the cases to a lower BIRADS level and resolved the need for closer follow-up.
ABSTRACT
BACKGROUND: Abbreviated magnetic resonance imaging (MRI) includes fewer sequences than standard MRI, which could be utilized for breast cancer detection. PURPOSE: To evaluate the diagnostic accuracy of abbreviated MRI protocol in screening and diagnostic settings. MATERIAL AND METHODS: All women with screening and diagnostic (problem-solving and preoperative staging) MRI examination were recruited from 2017 to 2020. Two expert radiologists assessed designed abbreviated protocol (fat-saturated T1-weighted [T1W] pre-contrast and two first fat-saturated T1W post-contrast series with reconstruction of their subtraction) including maximum intensity projection (MIP) and then evaluated standard protocol of breast MRI. Associated findings, including axillary lymphadenopathy and invasion to nipple, skin, or pectoralis muscle were also evaluated. The concordance rate of abbreviated with standard protocol in screening and diagnostic settings were also compared, based on BI-RADS classification. Diagnostic accuracy, sensitivity, specificity, and positive and negative predictive value were calculated. RESULTS: A total of 108 (26.5%) of 408 patients (mean age = 43 ± 9 years) were classified as BI-RADS 4-5 and considered positive findings based on suspicious enhancement (mass or non-mass enhancement). Compared to standard protocol, abbreviated protocol revealed >98% accuracy in the diagnostic setting as well as 100% accuracy in the screening setting. Concordance rates in screening and diagnostic settings were 99.6% and 98.1%, respectively. There was no discordance between abbreviated and standard protocol in the evaluation of associated findings. CONCLUSION: Abbreviated MRI protocol possesses substantial diagnostic accuracy in both screening and diagnostic settings. Additional information provided by standard protocol might not require for cancer detection.
Subject(s)
Breast Neoplasms , Breast , Female , Humans , Adult , Middle Aged , Sensitivity and Specificity , Breast/diagnostic imaging , Breast/pathology , Magnetic Resonance Imaging/methods , Predictive Value of Tests , Early Detection of Cancer/methods , Breast Neoplasms/diagnostic imaging , Breast Neoplasms/pathology , Retrospective StudiesABSTRACT
Introduction: Although neurologic involvement and neuroimaging abnormalities have been frequently identified in COVID-19 patients, the underlying factors remain unclear. In this study, we assessed the association of the neurological manifestations and neuroimaging features of hospitalized COVID-19 patients with their clinical, laboratory, and imaging characteristics. Methods: This multicenter cross-sectional study was conducted between September 2020 and March 2021 at two large academic hospitals in Tehran, Iran. We used census sampling from medical records to enroll hospitalized patients with a positive COVID-19 Polymerase chain reaction (PCR) test who underwent brain imaging due to presenting any acute neurologic symptom during hospital stay. Results: Of the 4372 hospitalized patients with COVID-19, only 211 met the inclusion criteria (35.5% with severe infection). Central nervous system and psychiatric manifestations were significantly more common in severe cases (p ≤ 0.044). Approximately, 30% had a new abnormality on their neuroimaging, with ischemic (38/63) and hemorrhagic (16/63) insults being the most common. The most frequent reasons that provoked cranial imaging were headache (27%), altered consciousness (25.6%), focal neurologic signs (19.9%), and delirium (18%). Analysis revealed a positive correlation for age, neutrophilia, lymphopenia, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) with the emergence of neuroimaging abnormalities (p ≤ 0.018). In addition, patients with new neuroimaging abnormalities had a significantly higher lung CT score than those without any pathologic findings (11.1 ± 4.8 vs. 5.9 ± 4.8, p < 0.001). Conclusion: Approximately 30% of the study population had various acute neuroimaging findings. The lung CT score, neutrophil count, and age were strong predictors of acute neuroimaging abnormalities in hospitalized COVID-19 patients.
ABSTRACT
BACKGROUND: This study aimed to investigate the potential relationship between diffusion kurtosis imaging (DKI)- derived parameters and lymphovascular space invasion (LVSI) in patients with cervical carcinoma. PATIENTS AND METHODS: This prospective study included 30 patients with cervical carcinoma. The patients underwent MRI, diffusion-weighted imaging (DWI), and DKI prior to surgery. The surgical pathology results were accepted as the reference standard for determining the LVSI status. The DKI-derived parameters, including mean diffusivity (MD) and mean kurtosis (MK), were measured. The apparent diffusion coefficient (ADC) value was also assessed. RESULTS: The MD value of LVSI positive cervical carcinomas was significantly lower than LVSI negative carcinomas (p-value = 0.01). MK value was significantly higher in LVSI positive tumors (p-value = 0.01). However, the ADC value did not show a significant difference between LVSI positive and LVSI negative tumors (p-value = 0.2). MD and MK parameters showed similar diagnostic accuracy in identifying the LVSI status, with the area under the curve of 0.77 and 0.78, respectively. CONCLUSION: In this study, DKI-derived parameters were associated with the LVSI status in cervical carcinomas. Further studies with larger sample size are required to confirm these results.
Subject(s)
Carcinoma , Uterine Cervical Neoplasms , Diffusion Magnetic Resonance Imaging/methods , Diffusion Tensor Imaging/methods , Female , Humans , Prospective Studies , Uterine Cervical Neoplasms/diagnostic imaging , Uterine Cervical Neoplasms/pathologyABSTRACT
Numerous solid breast masses require sophisticated analysis to establish a differential diagnosis. Consequently, complementary modalities such as ultrasound imaging are frequently required to evaluate mammographically further detected masses. Radiologists mentally integrate complementary information from images acquired of the same patient to make a more conclusive and effective diagnosis. However, it has always been a challenging task. This paper details a novel bimodal GoogLeNet-based CAD system that addresses the challenges associated with combining information from mammographic and sonographic images for solid breast mass classification. Each modality is initially trained using two distinct monomodal models in the proposed framework. Then, using the high-level feature maps extracted from both modalities, a bimodal model is trained. In order to fully exploit the BI-RADS descriptors, different image content representations of each mass are obtained and used as input images. In addition, using an ImageNet pre-trained GoogLeNet model, two publicly available databases, and our collected dataset, a two-step transfer learning strategy has been proposed. Our bimodal model achieves the best recognition results in terms of sensitivity, specificity, F1-score, Matthews Correlation Coefficient, area under the receiver operating characteristic curve, and accuracy metrics of 90.91%, 89.87%, 90.32%, 80.78%, 95.82%, and 90.38%, respectively. The promising results indicate that the proposed CAD system can facilitate bimodal suspicious mass analysis and thus contribute significantly to improving breast cancer diagnostic performance.
Subject(s)
Breast Neoplasms , Breast , Breast/diagnostic imaging , Breast Neoplasms/diagnostic imaging , Female , Humans , Machine Learning , Mammography/methods , ROC CurveABSTRACT
BACKGROUND: Triple-negative breast cancer (TNBC) is the most challenging subtype of breast cancer and does not benefit from the existing targeted therapies. In the present study, we used bioinformatics and experimental approaches to assess the genes that are somehow involved in the epithelial-mesenchymal transition (EMT) pathway which may explain the invasive features of TNBC. METHOD AND RESULTS: We analyzed five GEO datasets consisting of 657 breast tumors by GEO2R online software to achieve common differentially expressed genes (DEGs) between TNBC and non-TNBC tumors. The expression of the selected coding and non-coding genes was validated in 100 breast tumors, including fifty TNBC and fifty non-TNBC samples, using quantitative Real-Time PCR (qRT-PCR). The bioinformatics approach resulted in a final DEG list consisting of ten upregulated and seventeen downregulated genes (logFC ≥|1| and P < 0.05). Co-expression network construction indicated the FOXC1 transcription factor as a central hub node. Considering the notable role of FOXC1 in EMT, the expression levels of FOXC1-related lncRNAs, lnc-FOXCUT and lnc-DANCR, were also evaluated in the studied tumors. The results of qRT-PCR confirmed notable upregulation of FOXC1, lnc-FOXCUT, and lnc-DANCR in TNBC tissues compared to non-TNBC samples (P < 0.0001, P = 0.0005, and P = 0.0008, respectively). Moreover, ROC curve analysis revealed the potential biomarker role of FOXC1 in TNBC samples. CONCLUSION: Present study suggested that the deregulation of FOXC1/lnc-FOXCUT/lnc-DANCR axis may contribute to the aggressive features of triple-negative breast tumors. Therefore, this axis may be considered as a new probable therapeutic target in the treatment of TNBC.
Subject(s)
RNA, Long Noncoding , Triple Negative Breast Neoplasms , Cell Line, Tumor , Computational Biology , Forkhead Transcription Factors/genetics , Gene Expression Regulation, Neoplastic , Humans , RNA, Long Noncoding/genetics , RNA, Long Noncoding/metabolism , Triple Negative Breast Neoplasms/genetics , Triple Negative Breast Neoplasms/pathologyABSTRACT
OBJECTIVE: Breast ultrasound (BUS) is often performed as an adjunct to mammography in breast cancer screening or for evaluating breast lesions. Our aim was to design a practical and user-friendly format for BUS that could include the details of the Breast Imaging Reporting and Data System. MATERIALS AND METHODS: As a team of radiologists and surgeons trained in the management of breast diseases, we gathered and carried out the project in four phases-literature search and collection of present report formats, summarizing key points and preparing the first draft, seeking expert opinion and preparing the final format, and pilot testing-followed by a survey was answered by the research team's radiologists and surgeons. RESULTS: It produced a list of items to be stated in the BUS report, the final BUS report format, and the pilot format guide. Then, the radiologists used the format in three active ultrasound units in university-affiliated centers, and reports were referred to the surgeons. At the end of the project, the survey showed a high degree of ease of use, clarity, conciseness, comprehensiveness, and well-classified structure of the report format; but radiologists believed that the new organization took more time. CONCLUSION: We propose our design as a user-friendly and practical format for BUS reports. It should be used for a longer time and by various ultrasound centers in order to ascertain its benefits.
ABSTRACT
PURPOSE: To investigate the relationship between breast cancer imaging features on magnetic resonance imaging (MRI) and histopathological characteristics. METHODS AND MATERIALS: We prospectively enrolled 46 patients who underwent 1.5-T MRI with 68 breast malignant lesions from 2017 until 2019. Peritumoral edema was determined based on visual assessment on T2 weighted imaging. Lesions were categorized into two groups: A: with edema (48 lesions) and B: without edema (20 lesions). RESULTS: The tumor size was not different among two groups but multifocal-multicentric lesions were more common in the group A (70% vs. 35%). The axillary lymph nodes are most involved in group A. ER and PR positive lesions were more common in group B (90% vs. 56.3%) but in the group A, HER2 positive lesions were found to be more common (31.3% vs. 15%). The mean ADC value in tumors and peritumoral regions were lower (0.97 × 10-3 mm2/s, P = 0.023) and higher (1.85 × 10-3 mm2/s, P < 0.0001) in group A, respectively. Peritumoral ADC value was significantly higher in HER2-positive group. CONCLUSION: Breast carcinomas with peritumoral edema were found to be more multifocal-multicentric, with higher prevalence of axillary lymph node involvement, more HER 2-positive, with lower prevalence of ER/PR-positive, lower tumoral ADC and higher peritumoral ADC values.
Subject(s)
Breast Neoplasms , Biomarkers , Breast Neoplasms/diagnostic imaging , Diffusion Magnetic Resonance Imaging , Edema/diagnostic imaging , Humans , Retrospective StudiesABSTRACT
PURPOSE: Mammography plays a key role in the diagnosis of breast cancer; however, decision-making based on mammography reports is still challenging. This paper aims to addresses the challenges regarding decision-making based on mammography reports and propose a Clinical Decision Support System (CDSS) using data mining methods to help clinicians to interpret mammography reports. METHODS: For this purpose, 2441 mammography reports were collected from Imam Khomeini Hospital from March 21, 2018, to March 20, 2019. In the first step, these mammography reports are analyzed and program code is developed to transform the reports into a dataset. Then, the weight of every feature of the dataset is calculated. Random Forest, Naïve Bayes, K-nearest neighbor (K-NN), Deep Learning classifiers are applied to the dataset to build a model capable of predicting the need for referral to biopsy. Afterward, the models are evaluated using cross-validation with measuring Area Under Curve (AUC), accuracy, sensitivity, specificity indices. RESULTS: The mammography type (diagnostic or screening), mass and calcification features mentioned in the reports are the most important features for decision-making. Results reveal that the K-NN model is the most accurate and specific classifier with the accuracy and specificity values of 84.06% and 84.72% respectively. The Random Forest classifier has the best sensitivity and AUC with the sensitivity and AUC values of 87.74% and 0.905 respectively. CONCLUSIONS: Accordingly, data mining approaches are proved to be a helpful tool to make the final decision as to whether patients should be referred to biopsy or not based on mammography reports. The developed CDSS may also be helpful especially for less experienced radiologists.
ABSTRACT
Accurate electronic health records are important for clinical care, research, and patient safety assurance. Correction of misspelled words is required to ensure the correct interpretation of medical records. In the Persian language, the lack of automated misspelling detection and correction system is evident in the medicine and health care. In this article, we describe the development of an automated misspelling detection and correction system for radiology and ultrasound's free texts in the Persian language. To achieve our goal, we used n-gram language model and three different types of free texts related to abdominal and pelvic ultrasound, head and neck ultrasound, and breast ultrasound reports. Our system achieved the detection performance of up to 90.29% for radiology and ultrasound's free texts with the correction accuracy of 88.56%. Results indicated that high-quality spelling correction is possible in clinical reports. The system also achieved significant savings during the documentation process and final approval of the reports in the imaging department.
Subject(s)
Language , Natural Language Processing , Electronic Health Records , Female , Humans , Research Report , Ultrasonography, MammaryABSTRACT
PURPOSE: This work proposes a new reliable computer-aided diagnostic (CAD) system for the diagnosis of breast cancer from breast ultrasound (BUS) images. The system can be useful to reduce the number of biopsies and pathological tests, which are invasive, costly, and often unnecessary. METHODS: The proposed CAD system classifies breast tumors into benign and malignant classes using morphological and textural features extracted from breast ultrasound (BUS) images. The images are first preprocessed to enhance the edges and filter the speckles. The tumor is then segmented semiautomatically using the watershed method. Having the tumor contour, a set of 855 features including 21 shape-based, 810 contour-based, and 24 textural features are extracted from each tumor. Then, a Bayesian Automatic Relevance Detection (ARD) mechanism is used for computing the discrimination power of different features and dimensionality reduction. Finally, a logistic regression classifier computed the posterior probabilities of malignant vs benign tumors using the reduced set of features. RESULTS: A dataset of 104 BUS images of breast tumors, including 72 benign and 32 malignant tumors, was used for evaluation using an eightfold cross-validation. The algorithm outperformed six state-of-the-art methods for BUS image classification with large margins by achieving 97.12% accuracy, 93.75% sensitivity, and 98.61% specificity rates. CONCLUSIONS: Using ARD, the proposed CAD system selects five new features for breast tumor classification and outperforms state-of-the-art, making a reliable and complementary tool to help clinicians diagnose breast cancer.
ABSTRACT
Breast-region segmentation is an important step for density estimation and Computer-Aided Diagnosis (CAD) systems in Magnetic Resonance Imaging (MRI). Detection of breast-chest wall boundary is often a difficult task due to similarity between gray-level values of fibroglandular tissue and pectoral muscle. This paper proposes a robust breast-region segmentation method which is applicable for both complex cases with fibroglandular tissue connected to the pectoral muscle, and simple cases with high contrast boundaries. We present a decision-making framework based on geometric features and support vector machine (SVM) to classify breasts in two main groups, complex and simple. For complex cases, breast segmentation is done using a combination of intensity-based and atlas-based techniques; however, only intensity-based operation is employed for simple cases. A novel atlas-based method, that is called localized-atlas, accomplishes the processes of atlas construction and registration based on the region of interest (ROI). Atlas-based segmentation is performed by relying on the chest wall template. Our approach is validated using a dataset of 210 cases. Based on similarity between automatic and manual segmentation results, the proposed method achieves Dice similarity coefficient, Jaccard coefficient, total overlap, false negative, and false positive values of 96.3, 92.9, 97.4, 2.61 and 4.77%, respectively. The localization error of the breast-chest wall boundary is 1.97 mm, in terms of averaged deviation distance. The achieved results prove that the suggested framework performs the breast segmentation with negligible errors and efficient computational time for different breasts from the viewpoints of size, shape, and density pattern.