RESUMO
Being characterized by the self-adaption and high accuracy, the deep learning-based models have been widely applied in the 1D spectroscopy-related field. However, the "black-box" operation and "end-to-end" working style of the deep learning normally bring the low interpretability, where a reliable visualization is highly demanded. Although there are some well-developed visualization methods, such as Class Activation Mapping (CAM) and Gradient-weighted Class Activation Mapping (Grad-CAM), for the 2D image data, they cannot correctly reflect the weights of the model when being applied to the 1D spectral data, where the importance of position information is not considered. Here, aiming at the visualization of Convolutional Neural Network-based models toward the qualitative and quantitative analysis of 1D spectroscopy, we developed a novel visualization algorithm (1D Grad-CAM) to more accurately display the decision-making process of the CNN-based models. Different from the classical Grad-CAM, with the removal of the gradient averaging (GAP) and the ReLU operations, a significantly improved correlation between the gradient and the spectral location and a more comprehensive spectral feature capture were realized for 1D Grad-CAM. Furthermore, the introduction of difference (purity or linearity) and feature contribute in the CNN output in 1D Grad-CAM achieved a reliable evaluation of the qualitative accuracy and quantitative precision of CNN-based models. Facing the qualitative and adulteration quantitative analysis of vegetable oils by the combination of Raman spectroscopy and ResNet, the visualization by 1D Grad-CAM well reflected the origin of the high accuracy and precision brought by ResNet. In general, 1D Grad-CAM provides a clear vision about the judgment criterion of CNN and paves the way for CNN to a broad application in the field of 1D spectroscopy.
RESUMO
<p><b>OBJECTIVE</b>To evaluate the therapeutic efficacy of combined treatment with surgical decompression and drug for traumatic optic neuropathy (TON) and analyze the influential factors.</p><p><b>METHODS</b>A retrospective study on 69 patients (70 eyes) with TON treated with optic canal decompression through transnasal endoscopic approach and drug was conducted. The visual acuity was divided into 5 grades: no light perception (NLP), light perception (LP), hand move, count finger, > 0.02, marked as I-V respectively. Of 40 eyes with grade I, 18 eyes received emergency operation for severe optic canal fracture confirmed by CT; 22 eyes received corticosteroid therapy firstly and then operation. Of 30 eyes above grade I, 16 eyes with optic canal fracture confirmed by CT received emergency operation; 14 eyes received corticosteroid therapy firstly and 3 days later received operation. Postoperative follow-up lasted 3-12 months to observe the recovery of visual acuity.</p><p><b>RESULTS</b>The therapeutic efficacy of patients with the visual acuity of LP and above LP was better than that of NLP (90.0% to 27.5%), the difference had statistical significance (chi(2) = 26.98, P < 0.001). In operated group, the therapeutic efficacy in patients whose visual acuity was improved from NLP after glucocorticoid therapy (80.0%) was better than that of the patients with no improvement (5.9%), the difference had statistical significance (chi(2) = 12.09, P < 0.001).</p><p><b>CONCLUSION</b>The imaging findings of optic canal fracture can not be used as determinants for operation. The patients with NLP whose visual acuity had no improvement after corticosteroid therapy are poor candidates for surgical decompression. The visual acuity before treatment is the main factor affecting the therapeutic efficacy.</p>
