Computer-aided detection systems for breast masses: comparison of performances on full-field digital mammograms and digitized screen-film mammograms

RATIONALE AND OBJECTIVES: To compare the performance of computer aided detection (CAD) systems on pairs of full-field digital mammogram (FFDM) and screen-film mammogram (SFM) obtained from the same patients. MATERIALS AND METHODS: Our CAD systems on both modalities have similar architectures that consist of five steps. For FFDMs, the input raw image is first log-transformed and enhanced by a multiresolution preprocessing scheme. For digitized SFMs, the input image is smoothed and subsampled to a pixel size of 100 microm x 100 microm. For both CAD systems, the mammogram after preprocessing undergoes a gradient field analysis followed by clustering-based region growing to identify suspicious breast structures. Each of these structures is refined in a local segmentation process. Morphologic and texture features are then extracted from each detected structure, and trained rule-based and linear discriminant analysis classifiers are used to differentiate masses from normal tissues. Two datasets, one with masses and the other without masses, were collected. The mass dataset contained 131 cases with 131 biopsy proven masses, of which 27 were malignant and 104 benign. The true locations of the masses were identified by an experienced Mammography Quality Standards Act (MQSA) radiologist. The no-mass data set contained 98 cases. The time interval between the FFDM and the corresponding SFM was 0 to 118 days. RESULTS: Our CAD system achieved case-based sensitivities of 70%, 80%, and 90% at 0.9, 1.5, and 2.6 false positive (FP) marks/image, respectively, on FFDMs, and the same sensitivities at 1.0, 1.4, and 2.6 FP marks/image, respectively, on SFMs. CONCLUSIONS: The difference in the performances of our FFDM and SFM CAD systems did not achieve statistical significance.     

Computer-aided detection of colonic polyps using low-dose CT acquisitions

RATIONALE AND OBJECTIVES: This report proposes an alternative method for the automatic detection of colonic polyps that is robust enough to be directly applicable on low-dose computed tomographic data. MATERIALS AND METHODS: The polyp modeling process takes into account both the gray-level appearance of polyps (intensity profiles) and their geometry (extended Gaussian images). Spherical harmonic decompositions are used for comparison purposes, allowing fast estimation of the similarity between a candidate and a set of previously computed models. Starting from the original raw data (acquired at 55 mA), five patient data sets (prone and supine scans) are reconstructed at different dose levels (to 5 mA) by using different kernel filters, slice overlaps, and increments. Additionally, the efficacy of applying an edge-preserving smoothing filter before detection is assessed. RESULTS: Although image quality decreases when decreasing acquisition milliamperes, all polyps greater than 6 mm are detected successfully, even at 15 mA. Although not important at high doses, smoothing improves detection results for ultra-low-dose (tube current<15 mA) data. CONCLUSION: The advantage of low-dose scans is a significant decrease in effective dose from 4.93 to 1.61 mSv while retaining high detection values, particularly important when thinking of population screening.     

Computer-aided lung nodule detection in CT: results of large-scale observer test

RATIONALE AND OBJECTIVES: The objective is to study the incremental effects of using a computer-aided lung nodule detection (CAD) system on the performance of a large pool of observers. MATERIALS AND METHODS: A set of eight thin-section computed tomographic data sets with limited longitudinal coverage, containing a total of 22 lung nodules, was analyzed by using the automated nodule detection system. When applied to all eight cases, the CAD system alone achieved a detection rate of 86.4%, with 2.64 false-positive results per case. This study included 202 observers at a national radiology meeting: 39 thoracic radiologists, 95 non-thoracic radiologists, and 68 non-radiologists. Each participant read from one to eight cases in random order, first without and then with CAD system output available. Observer performance in nodule detection was measured before and after CAD was made available. Differences in performance of groups of observers before and after CAD were tabulated by mean, median, and SD in detection rate and number of false-positive results and tested by using nonparametric methods. RESULTS: In an analysis involving only the first randomly selected case read by all 202 participants, there were statistically significant increases in nodule detection rates and numbers of false-positive results for all types of observers. There was a significant difference in detection rates between radiologists and non-radiologists before CAD, but after CAD, there was no significant difference in detection rates between these observer types. In a second analysis involving 13 participants who read all eight cases, mean detection rates were 64.0% before CAD and 81.9% after CAD. Mean numbers of false-positive results were 0.144 per case before CAD and 0.173 after CAD. CONCLUSION: In a large observer study, use of a CAD system for nodule detection resulted in an incremental increase in detection rate, but also led to an increase in number of false-positive results. Also, CAD appears to be an equalizer of detection rates between observers of different levels of experience.     

Computer-aided detection of ischemic lesions related to subcortical vascular dementia on magnetic resonance images

RATIONALE AND OBJECTIVES: The purpose of this study was to develop an automated method for detection of the hyperintense ischemic lesions related to subcortical vascular dementia based on conventional magnetic resonance images (T1-weighted, T2-weighted, and fluid-attenuated inversion-recovery images [FLAIR]). MATERIALS AND METHODS: Our proposed method was based on subtraction between the T1-weighted image and the FLAIR image. First, a brain region was extracted by an automated thresholding technique based on a linear discriminant analysis for a pixel value histogram. Next, for enhancement of ischemic lesions, the T1-weighted image was subtracted from the fluid-attenuated inversion-recovery image. Ischemic lesion candidates were identified using a multiple gray-level thresholding technique and a feature-based region-growing technique on the subtraction image. Finally, an artificial neural network trained with 15 image features of the ischemic candidates was used to remove false-positives. We applied our method to nine patients with vascular dementia (age range, 64-94 years, mean age, 69.4 years; four males and five females), who were scanned on a 1.5-T magnetic resonance unit. RESULTS: Our method achieved a sensitivity of 90% with 4.0 false-positives per slice in detection of ischemic lesions. The overlap measure between ischemic lesion areas obtained by our method and a neuroradiologist was 60.7% on average. The ratio of ischemic lesion area to the whole brain area obtained by our method correlated with that determined by a neuroradiologist with a correlation coefficient of 0.911. CONCLUSION: Our preliminary results suggest that the proposed method may have feasibility for evaluation of the ischemic lesion area ratio.     

Automated mass detection in contrast-enhanced CT colonography: an approach based on contrast and volume

The purpose of this feasibility study was to design and test an algorithm for automating mass detection in contrast-enhanced CT colonography (CTC). Five patients with known colorectal masses underwent a pre-surgical contrast-enhanced (120 ml volume 1.6 g iodine/s injection rate, 60 s scan delay) CTC in high spatial resolution (16-slice CT: collimation: 16×0.75 mm, tablefeed: 24 mm/0.5 s, reconstruction increment: 0.5 mm). A CT-density- and volume-based algorithm searched for masses in the colonic wall, which was extracted before by segmenting and dilating the colonic air lumen and subtracting the inner air. A radiologist analyzed the detections and causes of false positives. All masses were detected, and false positives were easy to identify. Combining CT density with volume as a cut-off is a promising approach for automating mass detection that should be further refined and also tested in contrast-enhanced MR colonography. More information under .     

Computer-aided diagnosis scheme for detection of lacunar infarcts on MR images

RATIONALE AND OBJECTIVES: The detection and management of asymptomatic lacunar infarcts on magnetic resonance (MR) images are important tasks for radiologists to ensure the prevention of severe cerebral infarctions. However, accurate identification of the lacunar infarcts on MR images is a difficult task for the radiologists. Therefore the purpose of this study was to develop a computer-aided diagnosis scheme for the detection of lacunar infarcts to assist radiologists' interpretation as a "second opinion." MATERIALS AND METHODS: Our database comprised 1,143 T1- and 1,143 T2-weighted images obtained from 132 patients. The locations of the lacunar infarcts were determined by experienced neuroradiologists. We first segmented the cerebral region in a T1-weighted image by using a region growing technique for restricting the search area of lacunar infarcts. For identifying the initial lacunar infarcts candidates, a top-hat transform and multiple-phase binarization were then applied to the T2-weighted image within the segmented cerebral region. For eliminating the false positives (FPs), we determined 12 features--the locations x and y, signal intensity differences in the T1- and T2-weighted images, nodular components from a scale of 1 to 4, and nodular and linear components from a scale of 1 to 4. The nodular components and the linear components were obtained using a filter bank technique. The rule-based schemes and a support vector machine with 12 features were applied to the regions of the initial candidates for distinguishing between lacunar infarcts and FPs. RESULTS: Our computerized scheme was evaluated by using a holdout method. The sensitivity of the detection of lacunar infarcts was 96.8% (90/93) with 0.76 FP per image. CONCLUSIONS: Our computerized scheme would be useful in assisting radiologists for identifying lacunar infarcts in MR images.     

An improved quantitative measurement for thyroid cancer detection based on elastography

Objective

To evaluate color thyroid elastograms quantitatively and objectively.

Materials and methods

125 cases (56 malignant and 69 benign) were collected with the HITACHI Vision 900 system (Hitachi Medical System, Tokyo, Japan) and a liner-array-transducer of 6–13 MHz. Standard of reference was cytology (FNA—fine needle aspiration) or histology (core biopsy). The original color thyroid elastograms were transferred from red, green, blue (RGB) color space to hue, saturation, value (HSV) color space. Then, hard area ratio was defined. Finally, a SVM classifier was used to classify thyroid nodules into benign and malignant. The relation between the performance and hard threshold was fully investigated and studied.

Results

The classification accuracy changed with the hard threshold, and reached maximum (95.2%) at some values (from 144 to 152). It was higher than strain ratio (87.2%) and color score (83.2%). It was also higher than the one of our previous study (93.6%).

Conclusion

The hard area ratio is an important feature of elastogram, and appropriately selected hard threshold can improve classification accuracy.     

Comparison of the diagnostic performance of Magnetic Resonance Imaging (MRI), ultrasound and mammography for detection of breast cancer based on tumor type,breast density and patient's history: A review

ObjectiveBreast cancer is the most common malignancy in women. Mammography and ultrasound are commonly used in a clinical environment as the first choice for breast cancer detection. Magnetic Resonance Imaging (MRI) has been reported to reveal additional information. In the following review MRI, Ultrasound (US) and Mammography (MM) are all compared in terms of their diagnostic performance on breast cancer detection, depending on tumor type, breast density and patient's history.Key findingsEvaluating each modality alone, MRI provided an overall sensitivity and specificity of 94.6% (range 85.7%–100%) and 74.2% (range 25%–100%) respectively, while mammography showed that the overall sensitivity was at 54.5% (range 27%–86.8%) and specificity was 85.5% (range 62.9%–98.8%). The overall sensitivity and specificity of ultrasound was 67.2% (range 26.9%–87.5%) and 76.8% (range 18.8%–96.9%). When combining the results of all three techniques, it resulted in a sensitivity of 97.7% (range 95%–100%) and a specificity of 63.3% (range 37.1%–87.5%). In addition, contrast-enhanced mammography (CE-MM) and MRI (CE-MRI) illustrated an overall sensitivity and specificity for CE-MM was 90.5% (range 80.9%–100%) and 52.6% (range 15%–76.1%) and for CE-MRI, the overall sensitivity and specificity was 91.5% (range 89.1%–93.8%) and 64.7% (range 43.7%–85.7%).ConclusionAs modalities alone, the highest sensitivity has been observed for MRI and the lowest sensitivity for mammography regardless breast type, density, and history. Sensitivity is even more increased from the combination of US + MRI or MM + MRI or MRI + MM + US. The specificity seems to be affected by the size, type of the tumor and patient's history, however based on breast density, the highest specificity was observed by US alone.Implications for practiceBreast cancer screening is of outmost importance and identifying the best technique will improve cancer management. Combining techniques increases diagnostic ability compared with using modalities alone. CE-MM can be a viable option in dense breast tissue when there are contraindications to MRI as it also has high sensitivity based on the type of breast cancer.