Comparison of segmentation-free and segmentation-dependent computer-aided diagnosis of breast masses on a public mammography dataset.

PURPOSE: To compare machine learning methods for classifying mass lesions on mammography images that use predefined image features computed over lesion segmentations to those that leverage segmentation-free representation learning on a standard, public evaluation dataset. METHODS: We apply several classification algorithms to the public Curated Breast Imaging Subset of the Digital Database for Screening Mammography (CBIS-DDSM), in which each image contains a mass lesion. Segmentation-free representation learning techniques for classifying lesions as benign or malignant include both a Bag-of-Visual-Words (BoVW) method and a Convolutional Neural Network (CNN). We compare classification performance of these techniques to that obtained using two different segmentation-dependent approaches from the literature that rely on specific combinations of end classifiers (e.g. linear discriminant analysis, neural networks) and predefined features computed over the lesion segmentation (e.g. spiculation measure, morphological characteristics, intensity metrics). RESULTS: We report area under the receiver operating characteristic curve (AZ) values for malignancy classification on CBIS-DDSM for each technique. We find average AZ values of 0.73 for a segmentation-free BoVW method, 0.86 for a segmentation-free CNN method, 0.75 for a segmentation-dependent linear discriminant analysis of Rubber-Band Straightening Transform features, and 0.58 for a hybrid rule-based neural network classification using a small number of hand-designed features. CONCLUSIONS: We find that malignancy classification performance on the CBIS-DDSM dataset using segmentation-free BoVW features is comparable to that of the best segmentation-dependent methods we study, but also observe that a common segmentation-free CNN model substantially and significantly outperforms each of these (p < 0.05). These results reinforce recent findings suggesting that representation learning techniques such as BoVW and CNNs are advantageous for mammogram analysis because they do not require lesion segmentation, the quality and specific characteristics of which can vary substantially across datasets. We further observe that segmentation-dependent methods achieve performance levels on CBIS-DDSM inferior to those achieved on the original evaluation datasets reported in the literature. Each of these findings reinforces the need for standardization of datasets, segmentation techniques, and model implementations in performance assessments of automated classifiers for medical imaging.

Complete genomic landscape of a recurring sporadic parathyroid carcinoma.

Parathyroid carcinoma is a rare endocrine malignancy with an estimated incidence of less than 1 per million population. Excessive secretion of parathyroid hormone, extremely high serum calcium level, and the deleterious effects of hypercalcaemia are the clinical manifestations of the disease. Up to 60% of patients develop multiple disease recurrences and although long-term survival is possible with palliative surgery, permanent remission is rarely achieved. Molecular drivers of sporadic parathyroid carcinoma have remained largely unknown. Previous studies, mostly based on familial cases of the disease, suggested potential roles for the tumour suppressor MEN1 and proto-oncogene RET in benign parathyroid tumourigenesis, while the tumour suppressor HRPT2 and proto-oncogene CCND1 may also act as drivers in parathyroid cancer. Here, we report the complete genomic analysis of a sporadic and recurring parathyroid carcinoma. Mutational landscapes of the primary and recurrent tumour specimens were analysed using high-throughput sequencing technologies. Such molecular profiling allowed for identification of somatic mutations never previously identified in this malignancy. These included single nucleotide point mutations in well-characterized cancer genes such as mTOR, MLL2, CDKN2C, and PIK3CA. Comparison of acquired mutations in patient-matched primary and recurrent tumours revealed loss of PIK3CA activating mutation during the evolution of the tumour from the primary to the recurrence. Structural variations leading to gene fusions and regions of copy loss and gain were identified at a single-base resolution. Loss of the short arm of chromosome 1, along with somatic missense and truncating mutations in CDKN2C and THRAP3, respectively, provides new evidence for the potential role of these genes as tumour suppressors in parathyroid cancer. The key somatic mutations identified in this study can serve as novel diagnostic markers as well as therapeutic targets.

Genome-wide profiles of STAT1 DNA association using chromatin immunoprecipitation and massively parallel sequencing.

We developed a method, ChIP-sequencing (ChIP-seq), combining chromatin immunoprecipitation (ChIP) and massively parallel sequencing to identify mammalian DNA sequences bound by transcription factors in vivo. We used ChIP-seq to map STAT1 targets in interferon-gamma (IFN-gamma)-stimulated and unstimulated human HeLa S3 cells, and compared the method's performance to ChIP-PCR and to ChIP-chip for four chromosomes. By ChIP-seq, using 15.1 and 12.9 million uniquely mapped sequence reads, and an estimated false discovery rate of less than 0.001, we identified 41,582 and 11,004 putative STAT1-binding regions in stimulated and unstimulated cells, respectively. Of the 34 loci known to contain STAT1 interferon-responsive binding sites, ChIP-seq found 24 (71%). ChIP-seq targets were enriched in sequences similar to known STAT1 binding motifs. Comparisons with two ChIP-PCR data sets suggested that ChIP-seq sensitivity was between 70% and 92% and specificity was at least 95%.