Predicting lung cancer treatment response from CT images using deep learning

Association of Wnt-Inducible Signaling Pathway Protein 1 Genetic Polymorphisms With Lung Cancer Susceptibility and Platinum-Based Chemotherapy Response

Clinical decision making and research in hepatocellular carcinoma: Pivotal role of imaging techniques

Lung Cancer in today’s world is one of the major widespread dangerous diseases which is the subject of maximum deaths every year. Accurate detection of lung cancer could boost the endurance rate. Medical image processing has a significant impact on the recognition of lung tumors using Computer Tomography (CT) scan images. Images from a CT scan are widely used because they provide comprehensive imaging of tumor progression inside the lungs. Although different types of noise might be experienced while doing CT scans, producing it a monotonous task for recognizing tumors in the lung. Elimination of noise in CT images is a challenging task for medical diagnoses. The presence of noise in an image is inevitable. Hence reducing noise from the CT scan image is critical for further analysis. Hence various filtering techniques have been used that denoise and enhance the image and help in further evaluation of CT images for accurate lung cancer detection. This paper analyses the noises of different kinds in the CT images and different noise removal techniques which help in improving the accuracy of segmentation and feature extraction as they remove unwanted noise and contribute to the accurate detection of lung cancer. The various filtering methods are analyzed with salt along with pepper noise and speckle noise. The performances of different filters are computed in terms of metrics for evaluation like PSNR, SSIM, MSE, and SNR. The experimental results show that the median filter is more efficient in comparison to other filtering methods in eliminating noises that exist in lung CT images by owning fewer MSE values of 214.8522, high SNR value of 19.36304, SSIM value 0.595997, and high PSNR value of 24.80941.

e14572 Background: Objective response rates (ORR) utilizing RECIST or WHO criteria have been traditionally used as clinical endpoints to assess efficacy of cytotoxic chemotherapy. However in the era of immunotherapeutic agents, response rate assessment can be misleading due to nonconventional responses resulting in premature discontinuation of treatment. Wolchok et al proposed immune-related response criteria (irRC) to assess immune responses in melanoma. He also noted that stable disease which is not indicative of antitumor activity could be a potential surrogate marker of better clinical outcome. Brahmer et al reported nonconventional pattern of responses with Nivolumab in lung cancer. Methods: A retrospective chart review was done to assess differences in responses utilizing both RECIST v1.1 and irRC criteria in lung cancer patients treated with Nivolumab. Patients who received minimum of 4 cycles of Nivolumab were included in the study. The CT scans were reviewed by a senior radiologist. We reviewed 64 patients treated with Nivolumab between 4/30/2015 to 6/21/2016 and 30 patients were found eligible. Results: Patient characteristics. Male 53%; median age 61 years; squamous 47%; adeno 40%; small cell 13%; Median cycles of Nivolumab 6. Response rates utilizing both criteria was essentially similar with only 2 patients showing dis-concordant responses (6.6%). We utilized RECIST responses criteria for further analysis. Partial response (PR) 2/30; stable disease (SD) 12/30; progressive disease (PD) 15/30 and not evaluable (NE) 1/30. ORR 6.6%. Kaplan–Meier curve was used to analyze difference in survival between patients with PR+SD and PD. The median survival for PD was 9 months and PR+SD was not reached at median follow up of 1 year. Overall survival at 1 year for PD 33% and PR+SD 60% (p = 0.048). Hazard ratio for PD was 3.060 (p = 0.60). Conclusions: Patients with stable disease had a better survival and should be considered as responders though traditionally not included in the ORR. However irRC did not help in differentiating nonconventional immune responses in lung cancer patients. [Table: see text]

Introduction: The cancer burden is rising every year. Lung cancer is one of the most common cancers and non-small cell lung cancer is the most common type. Chemotherapy based on platinum drugs and third-generation nucleoside anti-metabolites such as gemcitabine are used widely. Gemcitabine has a complex metabolic pathway, with many mechanisms contributing to its cytotoxicity. Derangements in the metabolic pathway genes contribute to drug resistance and toxicity with this drug. Association studies including these genetic polymorphisms in the metabolic pathway, clinical outcomes, and cancer risk reported inter-individual differences. Thus, the aim of this study was to ascertain the role of these genetic variants in South Indian cancer patients treated with gemcitabine-based therapy.Methods: The study was done with 184 healthy volunteers for frequency establishment and 123 cancer patients were treated with gemcitabine-based chemotherapy for response and toxicity assessment. The participants were aged 18-65 years and resided in the southern states of India. DNA extraction was done from the leukocyte fraction of the blood by phenol-chloroform extraction procedures and genotyping was done by reverse transcription-polymerase chain reaction (RT-PCR) techniques to identify DNA repair gene polymorphisms. Tumor response was determined using Response evaluation criteria in solid tumors (RECIST) guidelines and toxicity using Common Terminology Criteria for Adverse Events (CTCAE), version 4.03. The patients were followed up for survival analysis.Results: The minor allele frequency of the single nucleotide polymorphism (SNP) NRF2-617 C>A (rs6721961) in the healthy population was 12.8%. SNPs were in Hardy-Weinberg equilibrium (p>0.05). Gender-based differences were not observed with the studied SNP in the healthy population and the lung cancer patients. These frequencies of NRF2 were found to be similar when compared to EUR (European) and all the South Asian subpopulations. They are significantly divergent compared to AFR (African), AMR (American), and EAS (East Asian) populations. The minor allele frequency in cancer patients was found to be 14.2% and the lung cancer risk with the SNP studied could not be detected. There was no association found with the response, toxicity, and survival among lung cancer patients.Conclusion: NRF2, being a multifaced molecule, did not show a significant association with lung cancer risk, response, and toxicity in patients with gemcitabine-based chemotherapy.

Detection of metastasis of mediastinal lymph nodes in lung cancer patients with an artificial intelligence model.

Chemotherapy of lung cancer has, until now, been an experimental approach that requires careful evaluation of tumor response. The growing number of lung cancer patients now undergoing chemotherapy has led to a rapid increase in the number of computed tomography (CT) scans performed. Eighty consecutive lung cancer patients (55 non-small cell and 25 small cell lung cancers) were included in a prospective study to analyze whether the standard chest roentgenography is as effective as computed tomography in evaluating tumor response. Both standard chest roentgenography and CT scanning were performed before the chemotherapy began and were repeated after 10 to 12 wk of treatment. Response evaluations were performed according to the World Health Organization recommendations. When two-dimensional measurements were possible, the indicator lesions were defined as measurable tumors. Both roentgenography methods were used, independently, to classify the response into the following categories: complete response, partial response, minor response, stable disease, and progressive disease. A comparison of CT scans versus standard chest roentgenography as a measurement of indicator lesion showed a concordance of borderline significance (kappa = 0.146, p less than 0.05); a significant asymmetry was demonstrated (McNemar = 35.6, p less than 0.001), indicating that CT scanning may be a more appropriate method for measuring tumors than standard chest roentgenography. Moreover, no concordance was observed comparing CT scan and standard chest roentgenography measurability in the subgroups of patients with T3 or T4 tumor, hilar tumor, and patients with pleural effusion or atelectasis in which the McNemar test of symmetry constantly showed a better measurability using CT scan.(ABSTRACT TRUNCATED AT 250 WORDS)

Plečiantis plaučių vėžio (PV) gydymo metodams ir galimybėms, kasdienėje klinikinėje praktikoje vis dažniau susiduriama su taikomų naviko atsako į gydymą vertinimo metodikų ribotumo ir tobulinimo klausimais. Standartizuota solidinių navikų atsako vertinimo metodika RECIST (angl. Response Evaluation Criteria in Solid Tumours) 1.1 ne visais atvejais gali užtikrinti ankstyvą progresuojančios ligos nustatymą. Tyrimo tikslas. Išanalizuoti išmatuojamų navikų matmenų ir tūrių pokyčių skirtumus, vertinant PV atsaką į gydymą kompiuterinės tomografijos (KT) vaizduose. Tyrimo metodai. Retrospektyviai įvertinti 2016–2019 m. įprastai atliktų KT tyrimų duomenys. Atrankos kriterijai: verifikuotas nesmulkių ląstelių PV, KT tyrimas atliktas prieš gydymą ir du KT tyrimai gydant; plaučiuose nustatytas bent vienas išmatuojamas navikinis darinys. Matuotas navikinio darinio ilgosios ašies matmuo (IA) ir tūris pagal elipsės (VT) ir sferos modelius (VS). Matavimai atlikti rankiniu ir pusiau automatiniu būdu, naudojant VitreaWorkstation™ CAD programinę įrangą. Dydžių pokyčiai vertinti remiantis RECIST 1.1 kategorijomis: progresuojanti liga (PL), dalinis atsakas ir stabili liga. Rezultatai. Tyrime analizuoti 125 pacientų duomenys. Išmatuoti 145 navikiniai dariniai ir įvertintas jų atsakas į gydymą. Nustatytas labai stiprus ryšys tarp IA ir VT prieš gydymą ir gydant (atitinkamai r=0,921, I r=0,937, II r=0,920). Vertinant VT pokyčius I ir II kontrolinių tyrimų vaizduose, PL nustatyta dažniau nei vertinant IA pokyčius, atitinkamai – 13,8 proc. ir 6,9 proc. Remiantis VT padidėjimu, I kontroliniame tyrime nustatytas PL dažnis 36,6 proc. (n=53) atitiko IA II kontroliniame tyrime nustatytą PL dažnį 35,9 proc. (n=52). Išvada. Nustatytas labai stiprus ryšys tarp naviko IA ir VT. Naviko VT padidėjimas gali būti jautresnis būdas įvertinti PL PV sergantiems pacientams.

Background: Current methods for assessing the response of lung cancer patients to therapy (e.g. RECIST criteria) lack consideration for mixed response (MR) where some lesions increase while others decrease simultaneously. MR poses a challenge in deciding the continuation of therapy. Understanding the clinical implications of MR on patient prognosis within radiographic response categories is crucial for establishing effective treatment strategies for NSCLC. Methods: This study analyzed 177 NSCLC patients with ≥2 lesions receiving first-line systemic treatment with immunotherapy. Clinical response was evaluated by RECIST v1.1 and immune-related RECIST (irRECIST). MR was defined as the simultaneous presence of ≥1 lesion increase and ≥1 lesion decrease during therapy. Two different definitions of MR were utilized. Definition-A (MR-A): the simultaneous presence of one or more lesions increasing ≥5mm and one or more lesions decreasing ≥5mm during therapy (Δ [baseline - first follow-up] ≥ 5 mm); and Definition-B (MR-B): the simultaneous presence of one or more lesions increasing >1mm and one or more lesions decreasing >1mm during therapy (any Δ [baseline - first follow-up]). Subgroup analysis was conducted using 94 patients with only intrathoracic lesions. Results: Among the 177 patients, tumor responses based on RECIST were 3 CR (1.7%),42 PR (23.7%), 87 SD (49.2%), and 45 PD (25.4%). Based on irRECIST, responses were 3 CR (1.7%), 46 PR (26.0%), 89 SD (50.3%), and 39 PD (22.0%). MR-A and MR-B occurred in 19.8% (n=35) and 45.2% (n=80) of cases, respectively. In the irRECIST PD group, MR-A demonstrated significantly different median progression-free survival (mPFS) and median overall survival (mOS) compared to non-MR-A (p<0.020, p<0.005, respectively). The mPFS for MR-A group was 1.4 months (range 0.5-52.9) and non-MR-A group was 5.5 months (range 0.2-97.2) with the hazard ratio being 2.91 (95% CI 1.13-7.51). The mOS for MR-A and non-MR-A groups were 4.8 months (range 0.7-61.5) and 11.8 months (range 0.2-97.2), respectively, hazard ratio 4.87 (95% CI 1.43-16.56). No other significant results were found for other RECIST and irRECIST categories. Subgroup analysis of only intrathoracic lesion patients revealed no statistically significant results. Conclusion: Contrary to our hypothesis, MR did not modify prognosis within the RECIST or irRECIST category. MR-A or MR-B in the RECIST or irRECIST PD group did not show a better prognosis, indicating the complexity of mixed responses in NSCLC patients treated with immunotherapy. Further research with larger cohorts are warranted. Citation Format: Seyoung Lee, Taegyu Um, Jeeyeon Lee, Peter Haseok Kim, Monica Yadav, Maria J. Chuchuca, Liam Il-Young Chung, Young Kwang Chae. Clinical implications of mixed response on prognosis of non-small cell lung cancer (NSCLC) patients treated with Immunotherapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3844.

The efficacy of treatment of cancer depends largely on early detection and correct prognosis. It is more important in case of pulmonary cancer, where the detection is based on identifying malignant nodules in the Computed Tomography (CT) scans of the lung. There are two problems for making correct decision about malignancy: (1) At early stage, the nodule size is small (length 5 to 10 mm). As the CT scan covers a volume of $30 {cm. }\times30 {cm. }\times40 {cm.}$, manually searching for nodules takes a very long time (approximately 10 minutes for an expert). (2) There are benign nodules and nodules due to other ailments like bronchitis, pneumonia, tuberculosis. To identify whether the nodule is carcinogenic needs long experience and expertise.In recent years, several works have been reported to classify lung cancer using not only the CT scan image, but also other features causing or related to cancer. In all recent works, for CT image analysis, 3-D Convolution Neural Network (CNN) is used to identify cancerous nodules. In spite of various preprocessing used to improve training efficiency, 3-D CNN is extremely slow. The aim of this work is to improve training efficiency by proposing a new deep NN model. It consists of a hierarchical (sliced) structure of recurrent neural network (RNN), where different layers of the hierarchy can be trained simultaneously, decreasing training time. In addition, selective attention (alignment) during training improves convergence rate. The result shows a 3-fold increase in training efficiency, compared to recent state-of-the-art work using 3-D CNN.

Achieving high feature reproducibility while preserving biological information is one of the main challenges for the generalizability of current radiomics studies. Non-clinical imaging variables, such as reconstruction kernels, have shown to significantly impact radiomics features. In this study, we retrain an open-source convolutional neural network (CNN) to harmonize computerized tomography (CT) images with various reconstruction kernels to improve feature reproducibility and radiomic model performance using epidermal growth factor receptor (EGFR) mutation prediction in lung cancer as a paradigm. In the training phase, the CNN was retrained and tested on 32 lung cancer patients’ CT images between two different groups of reconstruction kernels (smooth and sharp). In the validation phase, the retrained CNN was validated on an external cohort of 223 lung cancer patients’ CT images acquired using different CT scanners and kernels. The results showed that the retrained CNN could be successfully applied to external datasets with different CT scanner parameters, and harmonization of reconstruction kernels from sharp to smooth could significantly improve the performance of radiomics model in predicting EGFR mutation status in lung cancer. In conclusion, the CNN based method showed great potential in improving feature reproducibility and generalizability by harmonizing medical images with heterogeneous reconstruction kernels.

The objective of this study was to investigate the correlation between dual-energy computed tomography (DECT)-based iodine quantitation and fluorine-18 fluorodeoxyglucose (F-FDG) positron emission tomography (PET)/computed tomography (CT) imaging for response evaluation of lung cancers to treatment. In this prospective study, a total of 32 pairs of DECT and F-FDG PET/CT imaging acquired consecutively from 13 patients with primary or metastatic lung cancers receiving either radiotherapy alone or chemoradiotherapy were analyzed. Imaging examinations were performed before, immediately, and no later than 6 months after treatment for response evaluation. Iodine-related parameters including the total iodine uptake (TIU) and vital volume (VIV) from DECT and metabolic metrics such as the standardized uptake value normalized to lean body mass (SULpeak), metabolic tumor volume (MTV), and the total lesion glycolysis (TLG) from F-FDG-PET/CT were generated and measured by semiautomatic approaches. Dual-energy CT and PET/CT metrics were calculated and followed up with comparison with response evaluation criteria in solid tumors (RECIST). Analysis of pretreatment imaging data revealed a strong correlation between DECT metrics (RECIST, TIU, and VIV) and F-FDG PET/CT metrics (MTV, TLG) with coefficients of R ranging from 0.86 to 0.90 (P < 0.01). With the delivery of treatment, all measured DECT and PET/CT metrics significantly decreased whereas the descending amplitude in RECIST was significantly smaller than that of the remaining parameters (P < 0.05). During follow-up examinations, both metrics followed a similar changing pattern. Overall, strong consistency was found between RECIST, TIU, VIV and SULpeak, MTV, TLG (R covers 0.78-0.96, P < 0.05). Semiautomatic iodine-related quantitation in DECT correlated well with metabolism-based measurements in F-FDG PET/CT, suggesting that DECT-based iodine quantitation might be a feasible substitute for assessment of lung cancer response to chemoradiotherapy/radiotherapy with comparison with F-FDG PET/CT.

Lung cancer management challenges amidst COVID-19 pandemic: hope lives here.

Background: Lung cancer has the highest mortality rate in the world. Lung cancer management requires biomarkers to evaluate the chemotherapy response. Levels of Plasminogen Activator Inhibitor-1 (PAI-1) in malignancy &gt;14 ng/ml shows poor prognosis. Increased levels of PAI-1 are associated with stage, metastasis, and prognosis of lung cancer. The chemotherapy is thought to decrease PAI-1 levels. The aim this study is analyze the profile of PAI-1 levels before and after 3rd and 6th chemotherapy cycle and their changes by type of lung cancer and Response Evaluation Criteria in Solid Tumors (RECIST). Methods: This research was conducted from December 2016 to December 2017 at RSSA Malang. The study design was a cohort of 18 lung cancer patients. PAI-1 levels were measured by ELISA in 18 lung cancer stage III or IV patients before (PAI-1(1)) and after 3rd chemotherapy cycle (PAI-1(2)), and 9 patients after 6th chemotherapy cycle (PAI-1(3)). Average PAI-1 levels are presented in tables and graphs. Result: PAI-1(1) 2,151±0,564 ng/ml, PAI-1(2) 1,951±0,534 ng/ml, and PAI-1(3) 1,647±0,495 ng/ml. PAI-1 levels in non-small cell lung carcinoma (NSCLC) were 1.658±0.562 ng/ml and small cell lung carcinoma (SCLC) were 1.609±0.244 ng/ml. Levels of PAI-1(2) in partial response 1.784 ± 0.363 ng/ml, stable disease 1.980 ± 0.304 ng/ml, and progressive disease 2.020±0.635 ng/ml. Levels of PAI-1(3) in the partial response 1.427 ± 0.324 ng/ml and in progressive disease 2.085±0.532 ng/ml. Conclusions: PAI-1 levels in patients with stage III and IV lung cancer after chemotherapy were lower than before chemotherapy, and showed changes corresponding to the response according to RECIST. PAI-1 levels in NSCLC are higher than SCLC. (J Respir Indo 2018; 38(1): 48-56)

BackgroundChronic rhinosinusitis (CRS) is diagnosed with symptoms and objective endoscopy or computed tomography (CT). The Lund–Mackay score (LMS) is often used to determine the radiologic severity of CRS and make clinical decisions. This proof-of-concept study aimed to develop an automated algorithm combining a convolutional neural network (CNN) for sinus segmentation with post-processing to compute LMS directly from CT scans. ResultsRadiology Information System was queried for outpatient paranasal sinus CTs at a tertiary institution. We identified 1,399 CT scans which were manually labelled with LMS of individual sinuses. Seventy-seven CT scans with 13,668 coronal images were segmented manually for individual sinuses. Our model for segmentation achieved a mean Dice score of 0.85 for all sinus regions, except for the osteomeatal complex. For individual Dice scores were 0.95, 0.71, 0.78, 0.93, 0.86 for the maxillary, anterior ethmoid, posterior ethmoid, sphenoid, and frontal sinuses, respectively. LMS was computed automatically by applying adaptive image thresholding and pixel counting to the CNN’s segmented regions. A convolutional neural network (CNN) model was trained to segment each sinus region. Overall, the LMS model showed a high degree of accuracy with a score of 0.92, 0.99, 0.99, 0.97, 0.99, 0.86 for the maxillary, anterior ethmoid, posterior ethmoid, sphenoid, and frontal sinuses, respectively.ConclusionsReporting of paranasal sinus CT can be automated and potentially standardized with a CNN model to provide accurate Lund–Mackay score.