Weakly supervised deep learning for diagnosis of multiple vertebral compression fractures in CT.

OBJECTIVE: This study aims to develop a weakly supervised deep learning (DL) model for vertebral-level vertebral compression fracture (VCF) classification using image-level labelled data. METHODS: The training set included 815 patients with normal (n = 507, 62%) or VCFs (n = 308, 38%). Our proposed model was trained on image-level labelled data for vertebral-level classification. Another supervised DL model was trained with vertebral-level labelled data to compare the performance of the proposed model. RESULTS: The test set included 227 patients with normal (n = 117, 52%) or VCFs (n = 110, 48%). For a fair comparison of the two models, we compared sensitivities with the same specificities of the proposed model and the vertebral-level supervised model. The specificity for overall L1-L5 performance was 0.981. The proposed model may outperform the vertebral-level supervised model with sensitivities of 0.770 vs 0.705 (p = 0.080), respectively. For vertebral-level analysis, the specificities for each L1-L5 were 0.974, 0.973, 0.970, 0.991, and 0.995, respectively. The proposed model yielded the same or better sensitivity than the vertebral-level supervised model in L1 (0.750 vs 0.694, p = 0.480), L3 (0.793 vs 0.586, p < 0.05), L4 (0.833 vs 0.667, p = 0.480), and L5 (0.600 vs 0.600, p = 1.000), respectively. The proposed model showed lower sensitivity than the vertebral-level supervised model for L2, but there was no significant difference (0.775 vs 0.825, p = 0.617). CONCLUSIONS: The proposed model may have a comparable or better performance than the supervised model in vertebral-level VCF classification. CLINICAL RELEVANCE STATEMENT: Vertebral-level vertebral compression fracture classification aids in devising patient-specific treatment plans by identifying the precise vertebrae affected by compression fractures. KEY POINTS: • Our proposed weakly supervised method may have comparable or better performance than the supervised method for vertebral-level vertebral compression fracture classification. • The weakly supervised model could have classified cases with multiple vertebral compression fractures at the vertebral-level, even if the model was trained with image-level labels. • Our proposed method could help reduce radiologists' labour because it enables vertebral-level classification from image-level labels.

Development and validation of a hybrid deep learning-machine learning approach for severity assessment of COVID-19 and other pneumonias.

The Coronavirus Disease 2019 (COVID-19) is transitioning into the endemic phase. Nonetheless, it is crucial to remain mindful that pandemics related to infectious respiratory diseases (IRDs) can emerge unpredictably. Therefore, we aimed to develop and validate a severity assessment model for IRDs, including COVID-19, influenza, and novel influenza, using CT images on a multi-centre data set. Of the 805 COVID-19 patients collected from a single centre, 649 were used for training and 156 were used for internal validation (D1). Additionally, three external validation sets were obtained from 7 cohorts: 1138 patients with COVID-19 (D2), and 233 patients with influenza and novel influenza (D3). A hybrid model, referred to as Hybrid-DDM, was constructed by combining two deep learning models and a machine learning model. Across datasets D1, D2, and D3, the Hybrid-DDM exhibited significantly improved performance compared to the baseline model. The areas under the receiver operating curves (AUCs) were 0.830 versus 0.767 (p = 0.036) in D1, 0.801 versus 0.753 (p < 0.001) in D2, and 0.774 versus 0.668 (p < 0.001) in D3. This study indicates that the Hybrid-DDM model, trained using COVID-19 patient data, is effective and can also be applicable to patients with other types of viral pneumonia.

Importance of CT image normalization in radiomics analysis: prediction of 3-year recurrence-free survival in non-small cell lung cancer.

OBJECTIVES: To analyze whether CT image normalization can improve 3-year recurrence-free survival (RFS) prediction performance in patients with non-small cell lung cancer (NSCLC) relative to the use of unnormalized CT images. METHODS: A total of 106 patients with NSCLC were included in the training set. For each patient, 851 radiomic features were extracted from the normalized and the unnormalized CT images, respectively. After the feature selection, random forest models were constructed with selected radiomic features and clinical features. The models were then externally validated in the test set consisting of 79 patients with NSCLC. RESULTS: The model using normalized CT images yielded better performance than the model using unnormalized CT images (with an area under the receiver operating characteristic curve of 0.802 vs 0.702, p = 0.01), with the model performing especially well among patients with adenocarcinoma (with an area under the receiver operating characteristic curve of 0.880 vs 0.720, p < 0.01). CONCLUSIONS: CT image normalization may improve prediction performance among patients with NSCLC, especially for patients with adenocarcinoma. KEY POINTS: • After CT image normalization, more radiomic features were able to be identified. • Prognostic performance in patients was improved significantly after CT image normalization compared with before the CT image normalization. • The improvement in prognostic performance following CT image normalization was superior in patients with adenocarcinoma.

Correction of severe beam-hardening artifacts via a high-order linearization function using a prior-image-based parameter selection method.

PURPOSE: Polychromatic x-rays are used in most computed tomography scanners. In this case, a beam-hardening effect occurs, which degrades the image quality and distorts the shapes of objects in the reconstructed images. When the beam-hardening artifact is not severe, conventional correction methods can reduce the artifact reasonably well. However, highly dense materials, such as iron and titanium, can produce more severe beam-hardening artifacts, which often cannot be corrected by conventional methods. Moreover, when the size of the metal is large, severe darks bands due to photon starvation as well as beam-hardening are generated. The purpose of our study was to develop a new method for correcting severe beam-hardening artifacts and severe dark bands using a high-order polynomial correction function and a prior-image-based linearization method. METHODS: The initial estimate of an image free of beam-hardening (a prior image) was constructed from the initial reconstruction of the original projection data. Its corresponding beam-hardening-free projection data (a prior projection) were calculated by a projection operator onto the prior image. A new beam-hardening correction function G(praw ) with many high-order terms was effectively determined via a simple minimization process applied to the difference between the original projection data and the prior projection data. Using the determined correction function G(praw ), a corrected linearized sinogram pcorr can be obtained, which became effectively linear for the line integrals of the object. Final beam-hardening corrected images can be reconstructed from the linearized sinogram. The proposed method was evaluated in both simulation and real experimental studies. RESULTS: All investigated cases in both simulations and real experiments showed that the proposed method effectively removed not only streaks for moderate beam-hardening artifacts but also dark bands for severe beam-hardening artifacts without causing structural and contrast distortion. CONCLUSIONS: The prior-image-based linearization method exhibited better correction performance than conventional methods. Because the proposed method did not require time-consuming iterative reconstruction processes to obtain the optimal correction function, it can expedite the correction procedure and incorporate more high-order terms in the linearization correction function in comparison to the conventional methods.

Effects of H2 and electrochemical reducing power on metabolite production by Clostridium acetobutylicum KCTC1037.

A conventional fermenter (CF), a single-cathode fermenter (SCF), and a double-cathode fermenter (DCF) were employed to evaluate and compare the effects of H2 and electrochemical reducing power on metabolite production by Clostridium acetobutylicum KCTC1037. The source of the external reducing power for CF was H2, for the SCF was electrochemically reduced neutral red-modified graphite felt electrode (NR-GF), and for the DCF was electrochemically reduced combination of NR-GF and platinum plate electrodes (NR-GF/PtP). The metabolites produced from glucose or CO2 by strain KCTC1037 cultivated in the DCF were butyrate, ethanol, and butanol, but ethanol and butanol were not produced from glucose or CO2 by strain KCTC1037 cultivated in the CF and SCF. It is possible that electrochemically reduced NR-GF/PtP is a more effective source of internal and external reducing power than H2 or NR-GF for strain KCTC1037 to produce metabolites from glucose and CO2. This research might prove useful in developing fermentation technology to actualize direct bioalcohol production of fermentation bacteria from CO2.

Continuous bioelectricity production and sustainable wastewater treatment in a microbial fuel cell constructed with non-catalyzed granular graphite electrodes and permeable membrane.

Oxygen has been so far addressed as the most preferable terminal electron acceptor in the cathodes of microbial fuel cells (MFCs). However, to reduce the oxygen reduction overpotential at the cathode surface, eco-unfriendly and costly catalysts have been commonly employed. Here, we pursued the possibility of using a high surface area electrode to reduce the cathodic reaction overpotential rather than the utilization of catalyzed materials. A dual chambered MFC reactor was designed with the use of graphite-granule electrodes and a permeable membrane. The performance of the reactor in terms of electricity generation and organic removal rate was examined under a continuous-feed manner. Results showed that the maximum volumetric power of 4.4+/-0.2 W/m(3) net anodic compartment (NAC) was obtained at a current density of 11+/-0.5 A/m(3) NAC. The power output was improved by increasing the electrolyte ionic strength. An acceptable effluent quality was attained when the organic loading rate (OLR) of 2 kgCOD/m(3) NAC d was applied. The organic removal rate seemed to be less affected by shock loading. Our system can be suggested as a promising approach to make MFC-based technology economically viable for wastewater treatment applications. This study shows that current generation can be remarkably improved in comparison with several other studies using a low-surface-area plain graphite electrode.

Nitrifying biocathode enables effective electricity generation and sustainable wastewater treatment with microbial fuel cell.

Simultaneous organics removal and nitrification using a novel nitrifying biocathode microbial fuel cell (MFC) reactor were investigated in this study. Remarkably, the introduction of nitrifying biomass into the cathode chamber caused higher voltage outputs than that of MFC operated with the abiotic cathode. Results showed the maximum power density increased 18% when cathode was run under the biotic condition and fed by nitrifying medium with alkalinity/NH4+-N ratio of 8 (26 against 22 mW/m2). The voltage output was not differentiated when NH4+-N concentration was increased from 50 to 100 mg/L under such alkalinity/NH4+-N ratio. However, interestingly, the cell voltage rose significantly when the alkalinity/NH4+-N ratio was decreased to 6. Consequently, the maximum power density increased 68% in compared with the abiotic cathode MFC (37 against 22 mW/m2). Polarization curves demonstrated that both activation and concentration losses were lowered during the period of nitrifying biocathode operation. Ammonium was totally nitrified and mostly converted to nitrate in all cases of the biotic cathode conditions. High COD removal efficiency (98%) was achieved. In light of the results presented here, the application of nitrifying biocathode is not only able to integrate the nitrogen and carbon removal but also to enhance the power generation in MFC system. Our system can be suggested to open up a new feasible way for upgrading and retrofitting the existing wastewater treatment plant by the use of MFC-based technologies.

Simultaneous organics removal and bio-electrochemical denitrification in microbial fuel cells.

Simultaneous organics removal and bio-electrochemical denitrification using a microbial fuel cell (MFC) reactor were investigated in this study. The electrons produced as a result of the microbial oxidation of glucose in the anodic chamber were transferred to the anode, which then flowed to the cathode in the cathodic chamber through a wire, where microorganisms used the transferred electrons to reduce the nitrate. The highest power output obtained on the MFCs was 1.7 mW/m(2) at a current density of 15 mA/m(2). The maximum volumetric nitrate removal rate was 0.084 mg NO(3)(-)-N cm(-2) (electrode surface area) day(-1). The coulombic efficiency was about 7%, which demonstrated that a substantial fraction of substrate was lost without current generation.