Effect of hydrothermal carbonization on pyrolysis behavior, nutrients and metal species distribution in municipal sludge.

In this study, the effect of hydrothermal carbonation (HTC) on the pyrolysis behavior and the distribution of nutrients and metal species of waste-activated sludge (WAS) was investigated. Results showed that the pyrolysis activation energy range of WAS decreased from 11 to 57 kJ/mol to 10-36 kJ/mol when the hydrothermal carbonization was at 160 °C. As indicated by thermodynamic parameters, the hydrothermal carbonization process reduces the pyrolysis reaction activity of the hydrochar. The results of the chemical analysis indicate that hydrothermal carbonization significantly enhances the release of phosphorus and nitrogen, with maximum recovery at a temperature of 200 °C. The standard measurement and testing protocol revealed that hydrothermal carbonization increased the content of non-apatite inorganic P fraction in hydrochar and enhanced the availability of P. Heavy metal analysis shows that hydrothermal carbonization can strengthen the stability of heavy metals in WAS.

The Functions of Phasic Wing-Tip Folding on Flapping-Wing Aerodynamics.

Insects produce a variety of highly acrobatic maneuvers in flight owing to their ability to achieve various wing-stroke trajectories. Among them, beetles can quickly change their flight velocities and make agile turns. In this work, we report a newly discovered phasic wing-tip-folding phenomenon and its aerodynamic basis in beetles. The wings' flapping trajectories and aerodynamic forces of the tethered flying beetles were recorded simultaneously via motion capture cameras and a force sensor, respectively. The results verified that phasic active spanwise-folding and deployment (PASFD) can exist during flapping flight. The folding of the wing-tips of beetles significantly decreased aerodynamic forces without any changes in flapping frequency. Specifically, compared with no-folding-and-deployment wings, the lift and forward thrust generated by bilateral-folding-and-deployment wings reduced by 52.2% and 63.0%, respectively. Moreover, unilateral-folding-and-deployment flapping flight was found, which produced a lateral force (8.65 mN). Therefore, a micro-flapping-wing mechanism with PASFD was then designed, fabricated, and tested in a motion capture and force measurement system to validate its phasic folding functions and aerodynamic performance under different operating frequencies. The results successfully demonstrated a significant decrease in flight forces. This work provides valuable insights for the development of flapping-wing micro-air-vehicles with high maneuverability.

Neoadjuvant radioimmunotherapy in pancreatic cancer enhances effector T cell infiltration and shortens their distances to tumor cells.

Radiotherapy is hypothesized to have an immune-modulating effect on the tumor microenvironment (TME) of pancreatic ductal adenocarcinoma (PDAC) to sensitize it to anti-PD-1 antibody (a-PD-1) treatment. We collected paired pre- and posttreatment specimens from a clinical trial evaluating combination treatment with GVAX vaccine, a-PD-1, and stereotactic body radiation (SBRT) following chemotherapy for locally advanced PDACs (LAPC). With resected PDACs following different neoadjuvant therapies as comparisons, effector cells in PDACs were found to skew toward a more exhausted status in LAPCs following chemotherapy. The combination of GVAX/a-PD-1/SBRT drives TME to favor antitumor immune response including increased densities of GZMB+CD8+ T cells, TH1, and TH17, which are associated with longer survival, however increases immunosuppressive M2-like tumor-associated macrophages (TAMs). Adding SBRT to GVAX/a-PD-1 shortens the distances from PD-1+CD8+ T cells to tumor cells and to PD-L1+ myeloid cells, which portends prolonged survival. These findings have guided the design of next radioimmunotherapy studies by targeting M2-like TAM in PDACs.

Tumor-derived exosomal ADAM17 promotes pre-metastatic niche formation by enhancing vascular permeability in colorectal cancer.

BACKGROUND: Hematological metastasis has been recognized as a crucial factor contributing to the high rates of metastasis and mortality observed in colorectal cancer (CRC). Notably, exosomes derived from cancer cells participate in the formation of CRC pre-metastatic niches; however, the mechanisms underlying their effects are largely unknown. While our preliminary research revealed the role of exosome-derived disintegrin and metalloproteinase 17 (ADAM17) in the early stages of CRC metastasis, the role of exosomal ADAM17 in CRC hematogenous metastasis remains unclear. METHODS: In the present study, we isolated and purified exosomes using ultracentrifugation and identified exosomal proteins through quantitative mass spectrometry. In vitro, co-culture assays were conducted to evaluate the impact of exosomal ADAM17 on the permeability of the blood vessel endothelium. Vascular endothelial cell resistance, the cell index, membrane protein separation, flow cytometry, and immunofluorescence were employed to investigate the mechanisms underlying exosomal ADAM17-induced vascular permeability. Additionally, a mouse model was established to elucidate the role of exosomal ADAM17 in the modulation of blood vessel permeability and pre-metastatic niche formation in vivo. RESULTS: Our clinical data indicated that ADAM17 derived from the circulating exosomes of patients with CRC could serve as a blood-based biomarker for predicting metastasis. The CRC-derived exosomal ADAM17 targeted vascular endothelial cells, thus enhancing vascular permeability by influencing vascular endothelial cadherin cell membrane localization. Moreover, exosomal ADAM17 mediated the formation of a pre-metastatic niche in nude mice by inducing vascular leakage, thereby promoting CRC metastasis. Nonetheless, ADAM17 selective inhibitors effectively reduced CRC metastasis in vivo. CONCLUSIONS: Our results suggest that exosomal ADAM17 plays a pivotal role in the hematogenous metastasis of CRC. Thus, this protein may serve as a valuable blood-based biomarker and potential drug target for CRC metastasis intervention.

Precise quantification of microRNAs based on proximity ligation of AuNPs-immobilized DNA probes.

MiRNAs are critical regulators of target gene expression in many biological processes and are considered promising biomarkers for diseases. In this study, we developed a simple, specific, and sensitive miRNA detection method based on proximity ligation reaction, which is easy to operate. The method uses a pair of target-specific DNA probes immobilized on the same gold nanoparticles (AuNPs), which hybridize to the target miRNA. Hybridization brings the probes close together, allowing the formation of a continuous DNA sequence that can be amplified by Quantitative Real-time PCR (qPCR). This method eliminates the need for complex reverse transcription design and achieves high specificity for discriminating single base mismatches between miRNAs through a simple procedure. This method can sensitively measure three different miRNAs with a detection limit of 20 aM, providing high versatility and sensitivity, even distinguishing single-base variations among members of the miR-200 family with high selectivity. Due to its high selectivity and sensitivity, this method has important implications for the investigation of miRNA biological functions and related biomedical research.

Massive hemoptysis in pregnancy treated by ECMO combined with electronic bronchoscopy: A case report.

Background: Massive hemoptysis during pregnancy is very rare. Dieulafoy's disease is one of the causes of massive hemoptysis. There are few reports of ECMO use to treat massive hemoptysis during pregnancy. Findings: We report for the first time a patient with Dieulafoy's disease diagnosed at 29 weeks of pregnancy. The patient's hemoptysis occurred rapidly with large volumes. The bleeding amount reached 500 ml within half an hour, with the development of asphyxia and respiratory and cardiac arrest due to a blood clot blocking the airway. After successful cardiopulmonary resuscitation, the ventilator could not maintain effective ventilation. Emergency establishment of VV-ECMO was performed to maintain oxygen, and hemostasis was successfully achieved by performing bronchial artery embolization twice. We successfully cleaned blood clots in the airway four times by freezing and using a foreign body retrieval basket with an electronic bronchoscope. At the same time, small and smooth nodular lesions were found under bronchoscopy, and blood vessels with a diameter of 1.5 mm were found under Doppler mode with an ultrasonic bronchoscope, which was consistent with a diagnosis of Dieulafoy's disease. VV-ECMO was successfully stopped on the 3rd day of the disease course, tracheal intubation was successfully removed on the 5th day of the disease course, and the patient was discharged with no complications on the 16th day of the disease course.

Tumor- and Nerve-Derived Axon Guidance Molecule Promotes Pancreatic Ductal Adenocarcinoma Progression and Metastasis through Macrophage Reprogramming.

Axon guidance molecules were found to be the gene family most frequently altered in pancreatic ductal adenocarcinoma (PDA) through mutations and copy number changes. However, the exact molecular mechanism regarding PDA development remained unclear. Using genetically engineered mouse models to examine one of the axon guidance molecules, semaphorin 3D (SEMA3D), we found a dual role for tumor-derived SEMA3D in malignant transformation of pancreatic epithelial cells and a role for nerve-derived SEMA3D in PDA development. This was demonstrated by the pancreatic-specific knockout of the SEMA3D gene from the KRAS G12D and TP53 R 172 H mutation knock-in, PDX1-Cre (KPC) mouse model which demonstrated a delayed tumor initiation and growth comparing to the original KPC mouse model. Our results showed that SEMA3D knockout skews the macrophages in the pancreas away from M2 polarization, providing a potential mechanistic role of tumor-derived SEMA3D in PDA development. The KPC mice with the SEMA3D knockout remained metastasis-free, however, died from primary tumor growth. We then tested the hypothesis that a potential compensation mechanism could result from SEMA3D which is naturally expressed by the intratumoral nerves. Our study further revealed that nerve-derived SEMA3D does not reprogram macrophages directly, but reprograms macrophages indirectly through ARF6 signaling and lactate production in PDA tumor cells. SEMA3D increases tumor-secreted lactate which is sensed by GPCR132 on macrophages and subsequently stimulates pro-tumorigenic M2 polarization in vivo. Tumor intrinsic- and extrinsic-SEMA3D induced ARF6 signaling through its receptor Plexin D1 in a mutant KRAS-dependent manner. Consistently, RNA sequencing database analysis revealed an association of higher KRAS MUT expression with an increase in SEMA3D and ARF6 expression in human PDAs. Moreover, multiplex immunohistochemistry analysis showed an increased number of M2-polarized macrophages proximal to nerves in human PDA tissue expressing SEMA3D. Thus, this study suggests altered expression of SEMA3D in tumor cells lead to acquisition of cancer-promoting functions and the axon guidance signaling originating from nerves is "hijacked" by tumor cells to support their growth. Other axon guidance and neuronal development molecules may play a similar dual role which is worth further investigation. One sentence summary: Tumor- and nerve-derived SEMA3D promotes tumor progression and metastasis through macrophage reprogramming in the tumor microenvironment. STATEMENT OF SIGNIFICANCE: This study established the dual role of axon guidance molecule, SEMA3D, in the malignant transformation of pancreatic epithelial cells and of nerve-derived SEMA3D in PDA progression and metastasis. It revealed macrophage reprogramming as the mechanism underlying bothroles. Together, this research elucidated how inflammatory responses promote invasive PDA progression and metastasis through an oncogenic process.

Dual-inhibitory domain iCARs improve the efficiency of the AND-NOT gate CAR T strategy.

CAR (chimeric antigen receptor) T cell therapy has shown clinical success in treating hematological malignancies, but its treatment of solid tumors has been limited. One major challenge is on-target, off-tumor toxicity, where CAR T cells also damage normal tissues that express the targeted antigen. To reduce this detrimental side-effect, Boolean-logic gates like AND-NOT gates have utilized an inhibitory CAR (iCAR) to specifically curb CAR T cell activity at selected nonmalignant tissue sites. However, the strategy seems inefficient, requiring high levels of iCAR and its target antigen for inhibition. Using a TROP2-targeting iCAR with a single PD1 inhibitory domain to inhibit a CEACAM5-targeting CAR (CEACAR), we observed that the inefficiency was due to a kinetic delay in iCAR inhibition of cytotoxicity. To improve iCAR efficiency, we modified three features of the iCAR-the avidity, the affinity, and the intracellular signaling domains. Increasing the avidity but not the affinity of the iCAR led to significant reductions in the delay. iCARs containing twelve different inhibitory signaling domains were screened for improved inhibition, and three domains (BTLA, LAIR-1, and SIGLEC-9) each suppressed CAR T function but did not enhance inhibitory kinetics. When inhibitory domains of LAIR-1 or SIGLEC-9 were combined with PD-1 into a single dual-inhibitory domain iCAR (DiCARs) and tested with the CEACAR, inhibition efficiency improved as evidenced by a significant reduction in the inhibitory delay. These data indicate that a delicate balance between CAR and iCAR signaling strength and kinetics must be achieved to regulate AND-NOT gate CAR T cell selectivity.

Advances in the treatment of hepatogenous diabetes: A review.

Hepatogenous diabetes (HD) is a glycogen metabolism disorder that arises as a consequence of chronic liver disease. The condition is frequently detected in patients diagnosed with cirrhosis, which is a result of advanced liver disease. The prognosis for patients with HD is generally poor, and they are at a heightened risk for serious complications such as gastrointestinal bleeding, primary peritonitis, and hepatic encephalopathy. Hepatogenous diabetes progression is often associated with cirrhosis progression, which leads to the development of liver cancer and increased patient mortality. Despite the prevalence and severity of HD, no systematic treatment strategy for clinical management of the condition has been proposed by any research or institutions to date. This paper conducts an extensive review of recent advancements in HD treatment in the quest for an effective treatment approach that may improve the overall prognosis of HD.

FABNet: Frequency-Aware Binarized Network for Single Image Super-Resolution.

Remarkable achievements have been obtained with binary neural networks (BNN) in real-time and energy-efficient single-image super-resolution (SISR) methods. However, existing approaches often adopt the Sign function to quantize image features while ignoring the influence of image spatial frequency. We argue that we can minimize the quantization error by considering different spatial frequency components. To achieve this, we propose a frequency-aware binarized network (FABNet) for single image super-resolution. First, we leverage the wavelet transformation to decompose the features into low-frequency and high-frequency components and then employ a "divide-and-conquer" strategy to separately process them with well-designed binary network structures. Additionally, we introduce a dynamic binarization process that incorporates learned-threshold binarization during forward propagation and dynamic approximation during backward propagation, effectively addressing the diverse spatial frequency information. Compared to existing methods, our approach is effective in reducing quantization error and recovering image textures. Extensive experiments conducted on four benchmark datasets demonstrate that the proposed methods could surpass state-of-the-art approaches in terms of PSNR and visual quality with significantly reduced computational costs. Our codes are available at https://github.com/xrjiang527/FABNet-PyTorch.

Target Acquisition for Collimation System of Wireless Quantum Communication Networks in Low Visibility.

In severe low-visibility environments full of smoke, because of the performance degeneration of the near-infrared (NIR) collimation system of quantum drones communication networks, the improved dual-threshold method based on trend line analysis for long-wave infrared (LWIR) quantum cascade lasers (QCLs) is proposed, to achieve target acquisition. The simulation results show that smoke-scattering noise is a steeply varying medium-high-frequency modulation. At particle sizes less than 4 µm, the traditional dual-threshold method can effectively distinguish the target information from the smoke noise, which is the advantage of the LWIR laser compared to the NIR laser. For detecting lasers with high signal-to-noise ratios (SNRs), the method can achieve good target acquisition, by setting reasonable conventional thresholds, such as 0.7 times the peak intensity and 0.8 times the peak rising velocity. At low SNRs and steep intensity variation, the method can also achieve good target acquisition, by adaptively resetting new thresholds after filtering the detecting laser, such as 0.6 times the peak intensity and 0.6 times the peak rising velocity. The results of this paper will provide a reference for the performance improvement and refinement of the collimation system for wireless quantum communication networks in low visibility.

ADAM12 promotes the resistance of lung adenocarcinoma cells to EGFR-TKI and regulates the immune microenvironment by activating PI3K/Akt/mTOR and RAS signaling pathways.

Lung adenocarcinoma (LUAD) is a malignant respiratory disease, resulting in a heavy social burden. Epidermal growth factor receptor-tyrosine kinase inhibitor (EGFR-TKI) resistance and tumor immune microenvironment are important directions in the treatment of LUAD. In this study, we confirmed the role of ADAM metallopeptidase domain 12 (ADAM12) in LUAD development and progression. Our bioinformatic analysis was conducted to screen ADAM12 was correlated with EGFR-TKI and immune infiltration in LUAD patients. Our results showed that the transcription and post-transcription level of ADAM12 is significantly increased in tumor samples compared to normal samples, and ADAM12 correlated with poor prognosis in LUAD patients. High level of ADAM12 accelerated the LUAD progression via promoting proliferation, cell cycle, apoptosis escaping, immune escaping, EGFR-TKI resistance, angiogenesis, invasion and migration based on experiment validation in vitro and in vivo, which could be attenuated by ADAM12 knockdown. Further mechanistic studies suggested that the PI3K/Akt/mTOR and RAS signaling pathways were activated after ADAM12 knockdown. Therefore, ADAM12 might be validated as a possible molecular therapy target and prognostic marker for patients with LUAD.

Highly Enhancing CO2 Photoreduction by Metallization of an Imidazole-linked Robust Covalent Organic Framework.

Converting CO2 into value-added chemicals to solve the issues caused by carbon emission is promising but challenging. Herein, by embedding metal ions (Co2+ , Ni2+ , Cu2+ , and Zn2+ ) into an imidazole-linked robust photosensitive covalent organic framework (PyPor-COF), effective photocatalysts for CO2 conversion are rationally designed and constructed. Characterizations display that all of the metallized PyPor-COFs (M-PyPor-COFs) display remarkably high enhancement in their photochemical properties. Photocatalysis reactions reveal that the Co-metallized PyPor-COF (Co-PyPor-COF) achieves a CO production rate as high as up to 9645 µmol g-1 h-1 with a selectivity of 96.7% under light irradiation, which is more than 45 times higher than that of the metal-free PyPor-COF, while Ni-metallized PyPor-COF (Ni-PyPor-COF) can further tandem catalyze the generated CO to CH4 with a production rate of 463.2 µmol g-1 h-1 . Experimental analyses and theory calculations reveal that their remarkable performance enhancement on CO2 photoreduction should be attributed to the incorporated metal sites in the COF skeleton, which promotes the adsorption and activation of CO2 and the desorption of generated CO and even reduces the reaction energy barrier for the formation of different intermediates. This work demonstrates that by metallizing photoactive COFs, effective photocatalysts for CO2 conversion can be achieved.

A platform trial of neoadjuvant and adjuvant antitumor vaccination alone or in combination with PD-1 antagonist and CD137 agonist antibodies in patients with resectable pancreatic adenocarcinoma.

A neoadjuvant immunotherapy platform clinical trial allows for rapid evaluation of treatment-related changes in tumors and identifying targets to optimize treatment responses. We enrolled patients with resectable pancreatic adenocarcinoma into such a platform trial (NCT02451982) to receive pancreatic cancer GVAX vaccine with low-dose cyclophosphamide alone (Arm A; n = 16), with anti-PD-1 antibody nivolumab (Arm B; n = 14), and with both nivolumab and anti-CD137 agonist antibody urelumab (Arm C; n = 10), respectively. The primary endpoint for Arms A/B - treatment-related change in IL17A expression in vaccine-induced lymphoid aggregates - was previously published. Here, we report the primary endpoint for Arms B/C: treatment-related change in intratumoral CD8+ CD137+ cells and the secondary outcomes including safety, disease-free and overall survivals for all Arms. Treatment with GVAX+nivolumab+urelumab meets the primary endpoint by significantly increasing intratumoral CD8+ CD137+ cells (p = 0.003) compared to GVAX+Nivolumab. All treatments are well-tolerated. Median disease-free and overall survivals, respectively, are 13.90/14.98/33.51 and 23.59/27.01/35.55 months for Arms A/B/C. GVAX+nivolumab+urelumab demonstrates numerically-improved disease-free survival (HR = 0.55, p = 0.242; HR = 0.51, p = 0.173) and overall survival (HR = 0.59, p = 0.377; HR = 0.53, p = 0.279) compared to GVAX and GVAX+nivolumab, respectively, although not statistically significant due to small sample size. Therefore, neoadjuvant and adjuvant GVAX with PD-1 blockade and CD137 agonist antibody therapy is safe, increases intratumoral activated, cytotoxic T cells, and demonstrates a potentially promising efficacy signal in resectable pancreatic adenocarcinoma that warrants further study.

Immune repertoire sequencing reveals an abnormal adaptive immune system in COVID-19 survivors.

Accumulating evidence suggests that severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) impairs the adaptive immune system during acute infection. Still, it remains largely unclear whether the frequency and functions of T and B cells return to normal after the recovery of Coronavirus Disease 2019 (COVID-19). Here, we analyzed immune repertoires and SARS-CoV-2-specific neutralization antibodies in a prospective cohort of 40 COVID-19 survivors with a 6-month follow-up after hospital discharge. Immune repertoire sequencing revealed abnormal T- and B-cell expression and function with large T cell receptor/B cell receptor clones, decreased diversity, abnormal class-switch recombination, and somatic hypermutation. A decreased number of B cells but an increased proportion of CD19+ CD138+ B cells were found in COVID-19 survivors. The proportion of CD4+ T cells, especially circulating follicular helper T (cTfh) cells, was increased, whereas the frequency of CD3+ CD4- T cells was decreased. SARS-CoV-2-specific neutralization IgG and IgM antibodies were identified in all survivors, especially those recorded with severe COVID-19 who showed a higher inhibition rate of neutralization antibodies. All severe cases complained of more than one COVID-19 sequelae after 6 months of recovery. Overall, our findings indicate that SARS-CoV-2-specific antibodies remain detectable even after 6 months of recovery. Because of their abnormal adaptive immune system with a low number of CD3+ CD4- T cells and high susceptibility to infections, COVID-19 patients might need more time and medical care to fully recover from immune abnormalities and tissue damage.

Estimating Causal Effects using a Multi-task Deep Ensemble.

A number of methods have been proposed for causal effect estimation, yet few have demonstrated efficacy in handling data with complex structures, such as images. To fill this gap, we propose Causal Multi-task Deep Ensemble (CMDE), a novel framework that learns both shared and group-specific information from the study population. We provide proofs demonstrating equivalency of CDME to a multi-task Gaussian process (GP) with a coregionalization kernel a priori. Compared to multi-task GP, CMDE efficiently handles high-dimensional and multi-modal covariates and provides pointwise uncertainty estimates of causal effects. We evaluate our method across various types of datasets and tasks and find that CMDE outperforms state-of-the-art methods on a majority of these tasks.

Coupled Gold Nanoparticles with Aptamers Colorimetry for Detection of Amoxicillin in Human Breast Milk Based on Image Preprocessing and BP-ANN.

Antibiotic residues in breast milk can have an impact on the intestinal flora and health of babies. Amoxicillin, as one of the most used antibiotics, affects the abundance of some intestinal bacteria. In this study, we developed a convenient and rapid process that used a combination of colorimetric methods and artificial intelligence image preprocessing, and back propagation-artificial neural network (BP-ANN) analysis to detect amoxicillin in breast milk. The colorimetric method derived from the reaction of gold nanoparticles (AuNPs) was coupled to aptamers (ssDNA) with different concentrations of amoxicillin to produce different color results. The color image was captured by a portable image acquisition device, and image preprocessing was implemented in three steps: segmentation, filtering, and cropping. We decided on a range of detection from 0 µM to 3.9 µM based on the physiological concentration of amoxicillin in breast milk and the detection effect. The segmentation and filtering steps were conducted by Hough circle detection and Gaussian filtering, respectively. The segmented results were analyzed by linear regression and BP-ANN, and good linear correlations between the colorimetric image value and concentration of target amoxicillin were obtained. The R2 and MSE of the training set were 0.9551 and 0.0696, respectively, and those of the test set were 0.9276 and 0.1142, respectively. In prepared breast milk sample detection, the recoveries were 111.00%, 98.00%, and 100.20%, and RSDs were 6.42%, 4.27%, and 1.11%. The result suggests that the colorimetric process combined with artificial intelligence image preprocessing and BP-ANN provides an accurate, rapid, and convenient way to achieve the detection of amoxicillin in breast milk.

Highly Effective Generation of Singlet Oxygen by an Imidazole-Linked Robust Photosensitizing Covalent Organic Framework.

Developing effective photosensitizers to initiate the generation of singlet oxygen (1O2) is of great significance in both chemistry and physiology. Herein, linking the photoactive porphyrin moieties by in situ-formed robust imidazole groups, a covalent organic framework (COF), PyPor-COF, was successfully designed and synthesized. Detailed characterizations reveal that it not only possesses high crystallinity, permanent porosity, and robust stability but also shows a semiconductive photoresponse activity. As demonstrated by electron paramagnetic resonance experiments, the COF can initiate the generation of 1O2 efficiently under visible-light irradiation, the efficiency of which is higher than that of the pristine porphyrin-based reactant and even higher than some commonly used commercially available photosensitizing agents. Anticancer experiments prove that it can efficiently trigger the production of 1O2 in a physiological environment. This work demonstrates that the imidazole-linked porphyrin-incorporated COF is a highly promising photosensitizer that can even be applied in photodynamic therapy.

Artificial intelligence scale-invariant feature transform algorithm-based system to improve the calculation accuracy of Ki-67 index in invasive breast cancer: a multicenter retrospective study.

Background: Ki-67 is a key indicator of the proliferation activity of tumors. However, no standardized criterion has been established for Ki-67 index calculation. Scale-invariant feature transform (SIFT) algorithm can identify the robust invariant features to rotation, translation, scaling and linear intensity changes for matching and registration in computer vision. Thus, this study aimed to develop a SIFT-based computer-aided system for Ki-67 calculation in breast cancer. Methods: Hematoxylin and eosin (HE)-stained and Ki-67-stained slides were scanned and whole slide images (WSIs) were obtained. The regions of breast cancer (BC) tissues and non-BC tissues were labeled by experienced pathologists. All the labeled WSIs were randomly divided into the training set, verification set, and test set according to a fixed ratio of 7:2:1. The algorithm for identification of cancerous regions was developed by a ResNet network. The registration process between paired consecutive HE-stained WSIs and Ki-67-stained WSIs was based on a pyramid model using the feature matching method of SIFT. After registration, we counted the nuclear-stained Ki-67-positive cells in each identified invasive cancerous region using color deconvolution. To assess the accuracy, the AI-assisted result for each slice was compared with the manual diagnosis result of pathologists. If the difference of the two positive rate values is not greater than 10%, it was a consistent result; otherwise, it was an inconsistent result. Results: The accuracy of the AI-based algorithm in identifying breast cancer tissues in HE-stained slides was 93%, with an area under the curve (AUC) of 0.98. After registration, we succeeded in identifying Ki-67-positive cells among cancerous cells across the entire WSIs and calculated the Ki-67 index, with an accuracy rate of 91.5%, compared to the gold standard pathological reports. Using this system, it took about 1 hour to complete the evaluation of all the tested 771 pairs of HE- and Ki-67-stained slides. Each Ki-67 result took less than 2 seconds. Conclusions: Using a pyramid model and the SIFT feature matching method, we developed an AI-based automatic cancer identification and Ki-67 index calculation system, which could improve the accuracy of Ki-67 index calculation and make the data repeatable among different hospitals and centers.