Computer algorithms show potential for improving dermatologists' accuracy to diagnose cutaneous melanoma: Results of the International Skin Imaging Collaboration 2017.

BACKGROUND: Computer vision has promise in image-based cutaneous melanoma diagnosis but clinical utility is uncertain. OBJECTIVE: To determine if computer algorithms from an international melanoma detection challenge can improve dermatologists' accuracy in diagnosing melanoma. METHODS: In this cross-sectional study, we used 150 dermoscopy images (50 melanomas, 50 nevi, 50 seborrheic keratoses) from the test dataset of a melanoma detection challenge, along with algorithm results from 23 teams. Eight dermatologists and 9 dermatology residents classified dermoscopic lesion images in an online reader study and provided their confidence level. RESULTS: The top-ranked computer algorithm had an area under the receiver operating characteristic curve of 0.87, which was higher than that of the dermatologists (0.74) and residents (0.66) (P < .001 for all comparisons). At the dermatologists' overall sensitivity in classification of 76.0%, the algorithm had a superior specificity (85.0% vs. 72.6%, P = .001). Imputation of computer algorithm classifications into dermatologist evaluations with low confidence ratings (26.6% of evaluations) increased dermatologist sensitivity from 76.0% to 80.8% and specificity from 72.6% to 72.8%. LIMITATIONS: Artificial study setting lacking the full spectrum of skin lesions as well as clinical metadata. CONCLUSION: Accumulating evidence suggests that deep neural networks can classify skin images of melanoma and its benign mimickers with high accuracy and potentially improve human performance.

Results of the 2016 International Skin Imaging Collaboration International Symposium on Biomedical Imaging challenge: Comparison of the accuracy of computer algorithms to dermatologists for the diagnosis of melanoma from dermoscopic images.

BACKGROUND: Computer vision may aid in melanoma detection. OBJECTIVE: We sought to compare melanoma diagnostic accuracy of computer algorithms to dermatologists using dermoscopic images. METHODS: We conducted a cross-sectional study using 100 randomly selected dermoscopic images (50 melanomas, 44 nevi, and 6 lentigines) from an international computer vision melanoma challenge dataset (n = 379), along with individual algorithm results from 25 teams. We used 5 methods (nonlearned and machine learning) to combine individual automated predictions into "fusion" algorithms. In a companion study, 8 dermatologists classified the lesions in the 100 images as either benign or malignant. RESULTS: The average sensitivity and specificity of dermatologists in classification was 82% and 59%. At 82% sensitivity, dermatologist specificity was similar to the top challenge algorithm (59% vs. 62%, P = .68) but lower than the best-performing fusion algorithm (59% vs. 76%, P = .02). Receiver operating characteristic area of the top fusion algorithm was greater than the mean receiver operating characteristic area of dermatologists (0.86 vs. 0.71, P = .001). LIMITATIONS: The dataset lacked the full spectrum of skin lesions encountered in clinical practice, particularly banal lesions. Readers and algorithms were not provided clinical data (eg, age or lesion history/symptoms). Results obtained using our study design cannot be extrapolated to clinical practice. CONCLUSION: Deep learning computer vision systems classified melanoma dermoscopy images with accuracy that exceeded some but not all dermatologists.

When a private community neurology practice executes home sleep apnea testing: benefits identified and lessons learned in a retrospective observational study.

OBJECTIVE: To investigate the feasibility and utility of a private community neurology practice-initiated home sleep apnea testing (HSAT) program. METHODS: A private community neurology practice conducted HSAT on patients clinically identified as high risk for obstructive sleep apnea (OSA). An academic board-certified sleep specialist performed all study interpretations. The presence and severity of OSA and its association with patient demographics (eg, sex, age) and comorbid health conditions relevant to OSA were evaluated. RESULTS: During 2011-2014, 147 consecutive patients clinically identified as highly "at risk for OSA" during their neurological visit underwent HSAT. Sixty-one percent (n=89) of patients had a "positive" study with evidence of an apnea-hypopnea index of greater than 5 events per hour. Of those, 37% (n=54) had mild OSA and 24% (n=35) had moderate-severe OSA. OSA was more common among men (54%, n=48) and in individuals with a previous documented history of depression (33%, n=48) and hypertension 44% (n=64). OSA treatment was ordered in 44% (n=39) of patients by the neurologists or by a sleep specialist. Twenty-four percent (n=21) of all patients studied were referred to a sleep specialist. CONCLUSION: Implementation of HSAT in a (nonsleep) private community neurology practice in collaboration with an academic sleep program is recommended. Based on this observational study, community-based neurological practices and board-certified sleep specialists should consider teaming up to develop HSAT collaborative programs to open new sleep care access pathways for neurological patients often at risk for sleep apnea.

Connecting clinical aspects to corticomotor excitability in restless legs syndrome: a TMS study.

We assessed corticomotor excitability in the primary motor cortex (M1) of participants with moderate-to-severe restless legs syndrome (RLS) symptoms using transcranial magnetic stimulation (TMS) in relation to the clinical and sleep aspects of the disease. Thirty-five participants (20 F; mean age: 59.23 ± 1.66 years; range: 42-78 years) affected by primary RLS (off medications) and 31 age-matched controls (19 F; mean age: 57.90 ± 1.50 years; range: 43-79 years) underwent TMS following two nights of polysomnography (PSG). Paired-pulse TMS measures [short-interval intracortical inhibition (SICI), long-interval intracortical inhibition (LICI), and intracortical facilitation (ICF)] of the dominant M1hand and M1leg muscles were collected and analyzed in relation to clinical features of RLS and PSG. We found decreased corticomotor excitability in M1hand, whereas it was increased in M1leg, which was greater in patients with more severe RLS. Participants with RLS with a history of dopamine-agonist-induced symptom augmentation showed decreased LICI (reduced inhibition) compared to nonaugmented participants with RLS for M1leg. None of the TMS measures (M1hand or M1leg) correlated with the PSG parameters. This study shows hyperexcitability in M1leg, and this appears related to RLS disease severity and decreased excitability in M1hand. The results provide new insight into the complex neurobiology of RLS, particularly in more advanced stages of the disease.

Co-registration of magnetic resonance spectroscopy and transcranial magnetic stimulation.

Transcranial magnetic stimulation (TMS) is a widely used tool for noninvasive modulation of brain activity, that is thought to interact primarily with excitatory and inhibitory neurotransmitter systems. Neurotransmitters such as glutamate and GABA can be measured by magnetic resonance spectroscopy (MRS). An important prerequisite for studying the relationship between MRS neurotransmitter levels and responses to TMS is that both modalities should examine the same regions of brain tissue. However, co-registration of TMS and MRS has been little studied to date. This study reports on a procedure for the co-registration and co-visualization of MRS and TMS, successfully localizing the hand motor cortex, as subsequently determined by its functional identification using TMS. Sixteen healthy subjects took part in the study; in 14 of 16 subjects, the TMS determined location of motor activity intersected the (2.5cm)(3) voxel selected for MRS, centered on the so called 'hand knob' of the precentral gyrus. It is concluded that MRS voxels placed according to established anatomical landmarks in most cases agree well with functional determination of the motor cortex by TMS. Reasons for discrepancies are discussed.