Machine learning-based lung and colon cancer detection using deep feature extraction and ensemble learning

The detection of lung and colon cancer is a critical challenge in medical diagnosis, and machine learning (ML) and deep learning (DL) techniques are increasingly being used to enhance accuracy and efficiency. This review focuses on the integration of ML and DL methods for the combined detection of lung and colon cancer, emphasizing their strengths, limitations, and future potential. The motivation behind this study is to address the growing demand for accurate and early detection of these cancers, which significantly impacts treatment outcomes. Current methods often struggle with feature complexity, image variability, and computational intensity, which limit their real-world applicability. The aim is to consolidate various ML and DL techniques that have been employed for this purpose, highlighting how hybrid models can improve detection rates. The objective of this review is to provide a comprehensive analysis of different methodologies, their datasets, pre-processing techniques, feature extraction methods, and evaluation parameters. This review also explores recent advancements, such as transfer learning combined with fine-tuning techniques, which can further optimize performance in cancer detection. The findings suggest that while current methods show promise, further improvements in model generalization, interpretability, and computational efficiency are required to overcome existing limitations and expand clinical use.

To examine health care use and health care costs among partners of persons with cancer. Partners of patients with colon, rectal, lung, breast, and prostate cancer (N = 11,076) were identified via linked data from the Tumor Registry of Southern Sweden and Census Registry of Sweden. Health care use, total costs of health care, and diagnosis of the partner were studied before and after diagnosis of the cancer patient. Health care use for partners increased in terms of in-patient care after the cancer diagnosis. A significant increase was seen the second year for partners of patients with colon cancer (risk ratio [RR], 1.55; 95% CI, 1.28 to 1.87) and lung cancer (RR, 1.50; 95% CI, 1.26 to 1.79). Psychiatric diagnoses increased after the cancer diagnosis in the total sample, with a significant increase for partners of colon (RR, 2.66; 95% CI, 1.71 to 4.22), lung (RR, 3.16; 95% CI, 2.23 to 4.57), and prostate cancer patients (RR, 1.68; 95% CI, 1.32 to 2.15). Costs of care increased more than the consumer price index the two years after the cancer diagnosis. Costs of care increased most for male partners and especially for younger male partners (age 25 to 64 years) of patients with colon, rectal, and lung cancers. The results showed increased health care costs and an increase in psychiatric diagnoses after the cancer diagnosis among partners of cancer patients. Further research is needed to learn more about the situation of the partner and to identify persons at risk of psychiatric morbidity. Knowledge is also needed on how to support the partner in the most efficient way.

Introduction: Cardiovascular disease (CVD) and cancer share a common risk factor: chronic toxic stress/allostatic load (AL). Our previous research found that a 1-point increase in AL is linked to a 25-30% higher risk of major cardiac events (MACE) in prostate cancer patients. Hypothesis: Higher AL is associated with increased MACE risk in patients with breast, lung, and colon cancer. Methods: Patients ≥18 years diagnosed for the first time with the mentioned 3 cancers of interest between 2010-2019 at a large, hybrid academic-community practice were included in this retrospective cohort study. AL was modeled as an ordinal measure (0-11; Table 1 ). Adjusted Fine-Gray Cox proportional hazard regressions estimated the impact of AL pre- and 60 days post- cancer diagnosis on 2-year MACE. Results: Demographic characteristics are summarized in Table 1. Two-year mace MACE occurred in 12.3%, 23.3%, and 9.7% of breast, lung, and colon cancer patients, respectively. Heart failure was the most common MACE, observed in 7.9%, 14.4%, and 4.8% of breast, lung, and colon cancer patients, respectively. Before cancer diagnosis, a 1-point increase in AL raised MACE risk by 11% (aHR=1.11, 95% CI 1.07-1.16) in breast cancer, 18% (aHR=1.18, 95% CI 1.11-1.25) in lung cancer, and 9% (aHR=1.09, 95% CI 1.01-1.18) in colon cancer. After cancer diagnosis, a 1-point increase in AL increased MACE risk by 10% (aHR=1.10, 95% CI 1.06-1.15) in breast cancer and 19% (aHR=1.19, 95% CI 1.12-1.25) in lung cancer, but was not statistically significant in colon cancer (aHR=1.07, 95% CI 0.99-1.16). Conclusion(s): Higher AL levels prior to breast, lung, and colon cancer diagnoses are associated with increased MACE risk. Thus, effective implementation of patient-centered communication considering stress levels, stressors, and coping mechanisms is necessary during cardiovascular screening for these cancer patients. Further prospective studies are needed.

Adjuvant Chemotherapy Administration and Survival Outcomes in Lymphoma Survivors with Common Solid Tumors: A Population-Based Study

Differences in the average age at cancer diagnosis are observed across countries. We therefore aimed to assess international variation in the median age at diagnosis of common cancers worldwide, after adjusting for differences in population age structure. We used IARC's Cancer Incidence in Five Continents (CI5) Volume XI database, comprising cancer diagnoses between 2008 and 2012 from population-based cancer registries in 65 countries. We calculated crude median ages at diagnosis for lung, colon, breast and prostate cancers in each country, then adjusted for population age differences using indirect standardization. We showed that median ages at diagnosis changed by up to 10 years after standardization, typically increasing in low- and middle-income countries (LMICs) and decreasing in high-income countries (HICs), given relatively younger and older populations, respectively. After standardization, the range of ages at diagnosis was 12 years for lung cancer (median age 61-Bulgaria vs 73-Bahrain), 12 years for colon cancer (60-the Islamic Republic of Iran vs 72-Peru), 10 years for female breast cancer (49-Algeria, the Islamic Republic of Iran, Republic of Korea vs 59-USA and others) and 10 years for prostate cancer (65-USA, Lithuania vs 75-Philippines). Compared to HICs, populations in LMICs were diagnosed with colon cancer at younger ages but with prostate cancer at older ages (both pLMICS-vs-HICs < 0.001). In countries with higher smoking prevalence, lung cancers were diagnosed at younger ages in both women and men (both pcorr < 0.001). Female breast cancer tended to be diagnosed at younger ages in East Asia, the Middle East and Africa. Our findings suggest that the differences in median ages at cancer diagnosis worldwide likely reflect population-level variation in risk factors and cancer control measures, including screening.

A Comparative Study for Methodologies and Algorithms Used In Colon Cancer Diagnoses and Detection

Nowadays, a variety of cancers have become common among humans which unfortunately are the cause of death for many of these people. Early detection and diagnosis of cancers can have a significant impact on the survival of patients and treatment cost reduction. Colon cancer is the third and the second main cause of women's and men's death worldwide among cancers. Hence, many researchers have been trying to provide new methods for early diagnosis of colon cancer. In this study, we apply statistical hypothesis tests such as t-test and Mann–Whitney–Wilcoxon and machine learning methods such as Neural Network, KNN and Decision Tree to detect the most effective genes in the vital status of colon cancer patients. We normalize the dataset using a new two-step method. In the first step, the genes within each sample (patient) are normalized to have zero mean and unit variance. In the second step, normalization is done for each gene across the whole dataset. Analyzing the results shows that this normalization method is more efficient than the others and improves the overall performance of the research. Afterwards, we apply unsupervised learning methods to find the meaningful structures in colon cancer gene expressions. In this regard, the dimensionality of the dataset is reduced by employing Principle Component Analysis (PCA). Next, we cluster the patients according to the PCA extracted features. We then check the labeling results of unsupervised learning methods using different supervised learning algorithms. Finally, we determine genes which have major impact on colon cancer mortality rate in each cluster. Our conducted study is the first which suggests that the colon cancer patients can be categorized into two clusters. In each cluster, 20 effective genes were extracted which can be important for early diagnosis of colon cancer. Many of these genes have been identified for the first time.

To evaluate risk of early-onset dementia (EOD) after diagnosis of cancer among Medicaid beneficiaries. Longitudinal observational study of Medicaid enrollment, inpatient, and outpatient claims data from 26 states and Washington, DC, 2001-2019. Beneficiaries aged 18-64 with ≥6 months of enrollment were matched 1:1 on cancer status (lung, colon, breast, prostate) by age, sex, race, year and state. We estimated the weighted cumulative incidence functions of EOD at 1, 2, and 5 years after cancer diagnosis using the Aalen-Johansen estimator to account for the competing risk of death and cluster stratified analyses to account for matching. We calculated the corresponding risk differences (RD) and 95% confidence intervals (CI) using the 2.5th and 97.5th percentile of point estimates from 500 bootstrap resamples. The 5-year risk of EOD was 4.7% (95%CI: 4.5,5.0) and 4.7% (95%CI: 4.4, 4.9) among those with and without lung cancer, respectively (RD:0.08; 95%CI: -0.27,0.42). The 5-year risk of EOD was 4.1% (95%CI: 3.8, 4.4) and 3.9% (95%CI:3.7,4.3) among those with and without colon cancer, respectively, (RD 0.18; 95%CI: -0.25,0.55). The 5-year risk of EOD was 3.0% (95%CI: 2.8,3.1) and 2.9% (95%CI: 2.7,3.0) among those with and without breast cancer, respectively, (RD 0.10; 95%CI: -0.14,0.43). The 5-year risk of EOD was 4.6% (95%CI: 4.3,4.9) and 5.3% (95%CI: 4.9,5.7) among those with and without prostate cancer, respectively; those with prostate cancer had a lower EOD risk (RD -0.66; 95%CI: -1.2,-0.16). EOD incidence peaked at 4-5% among beneficiaries with and without cancer. Diagnosis of lung, colon, breast and prostate cancers were not strongly associated with EOD within 5 years. Additional work is needed to identify risk factors for EOD.

340P Prediction of relapse in colon cancer patients by machine learning models combining radiomics and deep features extracted from baseline computed tomography

Previous studies demonstrated that cancer sera contain antibodies, which react with a unique group of autologous cellular antigens called tumour-associated antigens (TAAs). This study determines whether a mini-array of multiple TAAs would enhance antibody detection and be a useful approach in colon cancer detection and diagnosis. The mini-array of multiple TAAs was composed of five TAAs including Imp1, p62, Koc, p53 and c-myc full-length recombinant proteins. Enzyme-linked immunosorbent assay (ELISA) was used to detect antibodies against these five TAAs in 46 sera from patients with colon cancer and also 58 sera from normal individuals. Antibody frequency to any individual TAA in colon cancer was variable and ranged from 15.2% to 23.9%. With the successive addition of TAAs to a final total of five antigens, there was a stepwise increase of positive antibody reactions reaching a sensitivity of 60.9% and a specificity of 89.7% in colon cancer. Positive and negative likelihood ratio was 5.91 and 0.43 respectively, which showed that the clinical diagnostic value of parallel assay of five TAAs was high. Positive and negative predictive values were respectively 82.4% and 74.3% indicating that parallel assay of five TAAs raised the diagnostic precision greatly. Agreement rate and Kappa value were 76.9% and 0.52 respectively, which indicated that the observed value of this assay had middle range coincidence with actual value. The data from this study further support our previous hypothesis that detection of autoantibodies for diagnosis of certain type of cancer can be enhanced by using a mini-array of several TAAs as target antigens. In 19 colon cancer sera with carcinoembryonic antigen (CEA) negative, 11 (57.9%) were found to have anti-TAA antibodies. When CEA and anti-TAAs were used together as markers in colon cancer detection, the diagnostic sensitivity could be raised from 60.9% to 82.6%. A customized antigen mini-array using a panel of appropriately selected TAAs can enhance autoantibody detection in immunodiagnosis of colon cancer. Anti-TAA and CEA were independent markers and the simultaneous use of these two markers significantly raised the sensitivity of colon cancer detection.

Artificial intelligence (AI) is revolutionizing early cancer detection by enhancing the sensitivity, efficiency, and precision of screening programs for breast, colorectal, and lung cancers. Deep learning algorithms, such as convolutional neural networks, are pivotal in improving diagnostic accuracy by identifying patterns in imaging data that may elude human radiologists. AI has shown remarkable advancements in breast cancer detection, including risk stratification and treatment planning, with models achieving high specificity and precision in identifying invasive ductal carcinoma. In colorectal cancer screening, AI-powered systems significantly enhance polyp detection rates during colonoscopies, optimizing the adenoma detection rate and improving diagnostic workflows. Similarly, low-dose CT scans integrated with AI algorithms are transforming lung cancer screening by increasing the sensitivity and specificity of early-stage cancer detection, while aiding in accurate lesion segmentation and classification. This review highlights the potential of AI to streamline cancer diagnosis and treatment by analyzing vast datasets and reducing diagnostic variability. Despite these advancements, challenges such as data standardization, model generalization, and integration into clinical workflows remain. Addressing these issues through collaborative research, enhanced dataset diversity, and improved explainability of AI models will be critical for widespread adoption. The findings underscore AI's potential to significantly impact patient outcomes and reduce cancer-related mortality, emphasizing the need for further validation and optimization in diverse healthcare settings.

Background/Objectives: Cancer screening programs are effective in the early detection of several cancers. This study aimed to identify the findings of cancer screenings that were conducted in 2016 and 2017.Methods/Statistical analysis: Sample were obtained from the National Health Insurance Service, and the cancer screening findings among both people with disabilities and those without were compared. Gastric, breast, colon, liver, and cervical cancer were analyze the findings of the screening programs. The findings were categorized as benign, suspicious of malignancy, or other categories (depending on type of cancer). Descriptive statistics and the X2-test were used to analyze the findings among the groups.Findings: The findings were as follows: For gastric cancer screenings, 0.30% in 2016 and 0.35% in 2017 of those screened who had brain injury were found to have gastric cancer, whereas 0.16% in 2016 and 0.15% in 2017 of those screened who were non-disabled where found to have gastric cancer (p<.001). For breast cancer screenings, 0.90% in 2016 and 0.54% in 2017 of those screened who had brain injury had findings that were suspicious of malignancy. For colon cancer screenings, 4.48% in 2016 of those screened who had brain injury were found to have colon cancer, whereas 3.70% in 2016 of those who were non-disabled were found to have colon cancer (p<.001). For liver cancer screenings, people with brain injury had the most findings that were suspicious of liver cancer in 2016 (p<.001), whereas people with physical impairment had the most findings that were suspicious of liver cancer in 2017 (p<.001). The non-disabled group had the highest incidence of carcinoma in situ among groups in 2016 (p<.001) and 2017 (p<.001).Improvements/Applications: Rigorous cancer screening programs should be implemented and expanded to include brain injury and physical impairment disabilities for the early detection of gastric, breast, colon, and liver cancer.

Cancer communication has begun to focus on prevention through lifestyle change. One possible consequence of this is that attitudes towards people with certain types of cancer could become increasingly negative. In this study, we assessed attributions of blame for five common cancers. Data were collected as part of a population-based survey of women's attitudes to cancer. A sample of 1,620 women rated how much they would blame someone who had a diagnosis of breast cancer, cervical cancer, bowel cancer, lung cancer, and leukemia, as well as chlamydia and obesity. Few women attributed any blame for someone with leukemia or breast cancer (9% and 15%). The proportions attributing some blame for bowel cancer and cervical cancer were slightly higher (23% and 37%). Patients with lung cancer were considered at least partly to blame by 70% of respondents, approaching chlamydia (87%) and obesity (96%). Attributions of blame for bowel, cervical, and lung cancer were most common among those with more qualifications, and women who knew that cervical cancer was linked to sexual activity were more likely to attribute blame to it (48% versus 32%). Cancers are not generally seen as a matter of personal responsibility, but lung cancer, and to a lesser extent, cervical and bowel cancer do attract blame attributions. Knowledge of the preventability of cancer seems to be associated with perceived responsibility. Increasing awareness of the preventable nature of some cancers might affect blame attributions. Additional research is needed to explore the implications of these findings.

Colonic adenocarcinoma is a major contributor to global mortality, highlighting the crucial need for efficient detection and classification techniques. This research presents a new method called XceptionTS for classifying and detecting colon cancer using colonoscopy pictures. The XceptionTS method utilizes deep transfer learning techniques by leveraging the Xception model architecture. Nonlinear Mean Filtering (NMF) is used as a noise reduction method in image processing to improve the quality of colonoscopy pictures. We combine the MobileNetV2 and ResNet-50 models for healthcare image segmentation and feature extraction, respectively. The XceptionTS classifier efficiently gives accurate class labels to medical photos by combining Tabu Search Optimization with the strong Xception architecture. The assessment of the effectiveness of XceptionTS model is done using a dataset of 1560 colonoscopy images. An extensive comparison study is undertaken by analyzing the efficacy of our suggested approach with existing research. The XceptionTS system outperforms previous methodologies in colon cancer classification and detection tasks, showing higher accuracy and robustness according to experimental results. Our findings indicate that the XceptionTS technique shows potential as an advanced tool to increase the effectiveness of Colonic adenocarcinoma diagnosis, which could lead to better patient outcomes and healthcare management.

P-0220 The Multiple Neoplasms Consist of Colon Cancer and Hematological Malignancy