RapidEELS: machine learning for denoising and classification in rapid acquisition electron energy loss spectroscopy.

Recent advances in detectors for imaging and spectroscopy have afforded in situ, rapid acquisition of hyperspectral data. While electron energy loss spectroscopy (EELS) data acquisition speeds with electron counting are regularly reaching 400 frames per second with near-zero read noise, signal to noise ratio (SNR) remains a challenge owing to fundamental counting statistics. In order to advance understanding of transient materials phenomena during rapid acquisition EELS, trustworthy analysis of noisy spectra must be demonstrated. In this study, we applied machine learning techniques to denoise high frame rate spectra, benchmarking with slower frame rate "ground truths". The results provide a foundation for reliable use of low SNR data acquired in rapid, in-situ spectroscopy experiments. Such a tool-set is a first step toward both automation in microscopy as well as use of these methods to interrogate otherwise poorly understood transformations.

Physical therapy management, surgical treatment, and patient-reported outcomes measures in a prospective observational cohort of patients with neurogenic thoracic outlet syndrome.

OBJECTIVE: To assess the results of physical therapy management and surgical treatment in a prospective observational cohort of patients with neurogenic thoracic outlet syndrome (NTOS) using patient-reported outcomes measures. METHODS: Of 183 new patient referrals from July 1 to December 31, 2015, 150 (82%) met the established clinical diagnostic criteria for NTOS. All patients underwent an initial 6-week physical therapy trial. Those with symptom improvement continued physical therapy, and the remainder underwent surgery (supraclavicular decompression with or without pectoralis minor tenotomy). Pretreatment factors and 7 patient-reported outcomes measures were compared between the physical therapy and surgery groups using t-tests and χ2 analyses. Follow-up results were assessed by changes in 11-item version of Disability of the Arm, Shoulder, and Hand (QuickDASH) scores and patient-rated outcomes. RESULTS: Of the 150 patients, 20 (13%) declined further treatment or follow-up, 40 (27%) obtained satisfactory improvement with physical therapy alone, and 90 (60%) underwent surgery. Slight differences were found between the physical therapy and surgery groups in the mean ± standard error degree of local tenderness to palpation (1.7 ± 0.1 vs 2.0 ± 0.1; P = .032), the number of positive clinical diagnostic criteria (9.0 ± 0.3 vs 10.1 ± 0.1; P = .001), Cervical-Brachial Symptom Questionnaire scores (68.0 ± 4.1 vs 78.0 ± 2.7; P = .045), and Short-Form 12-item physical quality-of-life scores (35.6 ± 1.5 vs 32.0 ± 0.8; P = .019) but not other pretreatment factors. During follow-up (median, 21.1 months for physical therapy and 12.0 months for surgery), the mean change in QuickDASH scores for physical therapy was -15.6 ± 3.0 (-29.5% ± 5.7%) compared with -29.8 ± 2.4 (-47.9% ± 3.6%) for surgery (P = .001). The patient-rated outcomes for surgery were excellent for 27%, good for 36%, fair for 26%, and poor for 11%, with a strong correlation between the percentage of decline in the QuickDASH score and patient-rated outcomes (P < .0001). CONCLUSIONS: The present study has demonstrated contemporary outcomes for physical therapy and surgery in a well-studied cohort of patients with NTOS, reinforcing that surgery can be effective when physical therapy is insufficient, even with substantial pretreatment disability. Substantial symptom improvement can be expected for ∼90% of patients after surgery for NTOS, with treatment outcomes accurately reflected by changes in QuickDASH scores. Within this cohort, it was difficult to identify specific predictive factors for individuals most likely to benefit from physical therapy alone vs surgery.