Effect of a novel segmentation algorithm on radiologists' diagnosis of breast masses using ultrasound imaging

A Brief Review on Breast Carcinoma and Deliberation on Current Non Invasive Imaging Techniques for Detection

BACKGROUND

Breast carcinoma is a life threatening disease that accounts for 25.1% of all carcinoma among women worldwide. Early detection of the disease enhances the chance for survival.

DISCUSSION

This paper presents comprehensive report on breast carcinoma disease and its modalities available for detection and diagnosis, as it delves into the screening and detection modalities with special focus placed on the non-invasive techniques and its recent advancement work done, as well as a proposal on a novel method for the application of early breast carcinoma detection.

CONCLUSION

This paper aims to serve as a foundation guidance for the reader to attain bird's eye understanding on breast carcinoma disease and its current non-invasive modalities.

Automated and real-time segmentation of suspicious breast masses using convolutional neural network

In this work, a computer-aided tool for detection was developed to segment breast masses from clinical ultrasound (US) scans. The underlying Multi U-net algorithm is based on convolutional neural networks. Under the Mayo Clinic Institutional Review Board protocol, a prospective study of the automatic segmentation of suspicious breast masses was performed. The cohort consisted of 258 female patients who were clinically identified with suspicious breast masses and underwent clinical US scan and breast biopsy. The computer-aided detection tool effectively segmented the breast masses, achieving a mean Dice coefficient of 0.82, a true positive fraction (TPF) of 0.84, and a false positive fraction (FPF) of 0.01. By avoiding positioning of an initial seed, the algorithm is able to segment images in real time (13–55 ms per image), and can have potential clinical applications. The algorithm is at par with a conventional seeded algorithm, which had a mean Dice coefficient of 0.84 and performs significantly better (P< 0.0001) than the original U-net algorithm.

The diagnostic performance of leak-plugging automated segmentation versus manual tracing of breast lesions on ultrasound images

PURPOSE

To assess the diagnostic performance of a leak-plugging segmentation method that we have developed for delineating breast masses on ultrasound images.

MATERIALS AND METHODS

Fifty-two biopsy-proven breast lesion images were analyzed by three observers using the leak-plugging and manual segmentation methods. From each segmentation method, grayscale and morphological features were extracted and classified as malignant or benign by logistic regression analysis. The performance of leak-plugging and manual segmentations was compared by: size of the lesion, overlap area (O_a ) between the margins, and area under the ROC curves (A_z ).

RESULTS

The lesion size from leak-plugging segmentation correlated closely with that from manual tracing (R² of 0.91). O_a was higher for leak plugging, 0.92 ± 0.01 and 0.86 ± 0.06 for benign and malignant masses, respectively, compared to 0.80 ± 0.04 and 0.73 ± 0.02 for manual tracings. Overall O_a between leak-plugging and manual segmentations was 0.79 ± 0.14 for benign and 0.73 ± 0.14 for malignant lesions. A_z for leak plugging was consistently higher (0.910 ± 0.003) compared to 0.888 ± 0.012 for manual tracings. The coefficient of variation of A_z between three observers was 0.29% for leak plugging compared to 1.3% for manual tracings.

CONCLUSION

The diagnostic performance, size measurements, and observer variability for automated leak-plugging segmentations were either comparable to or better than those of manual tracings.

Segmentation of Breast Lesions in Ultrasound Images through Multiresolution Analysis Using Undecimated Discrete Wavelet Transform

Earliest detection and diagnosis of breast cancer reduces mortality rate of patients by increasing the treatment options. A novel method for the segmentation of breast ultrasound images is proposed in this work. The proposed method utilizes undecimated discrete wavelet transform to perform multiresolution analysis of the input ultrasound image. As the resolution level increases, although the effect of noise reduces, the details of the image also dilute. The appropriate resolution level, which contains essential details of the tumor, is automatically selected through mean structural similarity. The feature vector for each pixel is constructed by sampling intra-resolution and inter-resolution data of the image. The dimensionality of feature vectors is reduced by using principal components analysis. The reduced set of feature vectors is segmented into two disjoint clusters using spatial regularized fuzzy c-means algorithm. The proposed algorithm is evaluated by using four validation metrics on a breast ultrasound database of 150 images including 90 benign and 60 malignant cases. The algorithm produced significantly better segmentation results (Dice _coef = 0.8595, boundary displacement error = 9.796, d_vi = 1.744, and global consistency error = 0.1835) than the other three state of the art methods.

Assessment of combined textural and morphological features for diagnosis of breast masses in ultrasound

The objective of this study is to assess the combined performance of textural and morphological features for the detection and diagnosis of breast masses in ultrasound images. We have extracted a total of forty four features using textural and morphological techniques. Support vector machine (SVM) classifier is used to discriminate the tumors into benign or malignant. The performance of individual as well as combined features are assessed using accuracy(Ac), sensitivity(Se), specificity(Sp), Matthews correlation coefficient(MCC) and area AZ under receiver operating characteristics curve. The individual features produced classification accuracy in the range of 61.66% and 90.83% and when features from each category are combined, the accuracy is improved in the range of 79.16% and 95.83%. Moreover, the combination of gray level co-occurrence matrix (GLCM) and ratio of perimeters (P ratio ) presented highest performance among all feature combinations (Ac 95.85%, Se 96%, Sp 91.46%, MCC 0.9146 and AZ 0.9444).The results indicated that the discrimination performance of a computer aided breast cancer diagnosis system increases when textural and morphological features are combined.

A computer-aided diagnostic algorithm improves the accuracy of transesophageal echocardiography for left atrial thrombi: a single-center prospective study

OBJECTIVES

We investigated whether transesophageal echocardiography (TEE) assisted with a computer-aided diagnostic (CAD) algorithm was superior to TEE in diagnosing left atrial (LA)/left atrial appendage (LAA) thrombi in patients with atrial fibrillation (AF) in a single prospective study.

METHODS

Transesophageal echocardiography was performed in patients with AF, and images were reconstructed. Gray level co-occurrence matrix-based features were calculated and then classified using an artificial neural network. The original data and processed images by the CAD system were studied by 5 radiologists independently in a blind manner. The diagnostic performance of each radiologist was evaluated.

RESULTS

One hundred thirty patients with AF were investigated. Thirty-one patients (23.9%) had a diagnosis of LA/LAA thrombi. The mean sensitivity ± SD of TEE for LA/LAA thrombi was 0.933 ± 0.027, which was noticeably improved by CAD (0.955 ± 0.021; P < .05). The specificity of TEE was 0.811 ± 0.055, which was markedly lower than that by TEE plus CAD (0.970 ± 0.009; P < .05). The positive predictive value of TEE was low (0.613 ± 0.073) compared to that of TEE plus CAD (0.908 ± 0.027; P < .001), whereas the negative predictive values were comparable for TEE, CAD, and TEE plus CAD. Diagnosis of an LA/LAA thrombus by TEE plus CAD had a higher accuracy rate (0.966 ± 0.011) than that by TEE (0.840 ± 0.047; P < .01). The mean area under the receiver operating characteristic curve (Az) for TEE was 0.834 ± 0.009 (95% confidence interval [CI], 0.815-0.852), which was markedly lower than the Az for TEE plus CAD (0.932 ± 0.005; 95% CI, 0.921-0.943). The use of CAD significantly improved the Az values for all 5 radiologists (P < .001).

CONCLUSIONS

The CAD algorithm significantly improves the diagnostic accuracy of TEE for LA/LAA thrombi in patients with AF.