Abstract:
Cancer is one of the deadliest diseases so far found in the human beings. During the years 2012 and 2014, the death rate due to this deadly disease increased by nearly 6%. In 2012, there were 478,180 losses out of 2,934,314 cases stated. In 2013 there were 465,169 demises out of 3,016,628 cases. In 2014, 491,598 people deceased in out of 2,820,179 cases. In 2017, 9.6 million people are projected to have died from the various types of cancer. Every sixth bereavement in the world is due to cancer, making it the second prominent cause of deaths. Such a disease has its beginning with a small lump in any of the body parts, this is known to be a tumor. We cannot claim all tumors to be cancer tumors but most of the cancers begin with a tumor. So we are proposing a model to classify the tumor occurring in human as normal ones or cancer tumors, once they are categorized as Cancer Tumors, the type of cancer is predicted and the stage and immediate precautions are also prescribed.
Problem identification:
In this drastically changing lifestyle of people, there are many possibilities of cancer occurring, this disease occurs due to varied reasons like, organic or general human factors, such as age, gender, inherited genetic defects and skin type, environmental contact, for instance to radon and UV radiation, and fine particulate matter, job related risk factors, including consumption of carcinogens such as many chemicals, radioactive materials and asbestos and last but not least lifestyle-related factors such as simple food which we consume day to day as all the fruits and vegetables we eat has loads and loads of chemicals in the form of fertilizers and pesticides and the all junk food we consume leads us in the intake of cancer causing agents, so in these inevitable conditions every one of us irrespective of the life we live has a chance for this disease, but there are cases where people have survived from this, these are due to early detection and diagnosis so we aim at detecting this disease at an earlier stage with its early symptoms like tumor markers.
Proposed solution:
Tests for cancers are vast so one of the tests can be done with the use of tumor markers. The tumor image is captured in CT scan or MRI scan these kind of X- Ray images the unique features or tumor markers corresponding to cancer and its type are detected if any by our proposed model. The features from the already available patient’s datasets are extracted and stored; both the features are extracted and sent to the model for comparison. The type of cancer is predicted if the tumor is confirmed to be cancerous, we here can use two types of datasets both numerical and textual ones and image datasets our model should be flexible to use both and manipulate them with higher accuracy.
How technologies contribute to it:
For the above suggested solution we need many machine learning libraries like pandas, scikitlearn, scipy, numpy, matplotlib, keras, tensorflow, opencv, knn, regression, recommender systems and lot more. For instance to read all the data (numeric) and to explore features we need numpy and pandas, to plot features extracted from opencv or pandas exploration in graphs or visual means we need matplotlib, to analyze images, layering for neural networks, extract, compare, match feature points and come to fruitful conclusions we require keras, tensorflow, opencv libraries and to finally predict the outcomes we use knn classification, regression prediction techniques and finally we use recommender systems to provide suggestions for the problem faced by which the situation can be kept under control. Artificial intelligence is used in this because we make the model to learn the features, changes and all the factors some by supervised learning and some using unsupervised learning and out if that knowledge we gain future predictions to control the factors and it is impossible by human to interpret certain digital images and analyze voluminous datasets hence we use machine learning and artificial intelligence models
Improvements suggested:
The future enhancements of this proposed model are, this can be even modeled into a mobile application and using machine learning algorithms and the collected datasets we can predict for each individual to be tested for cancer or advises on lifestyle changes can be given, also the prediction of the amount of cancer causing agents in food are calculated and change in food habits can be suggested and healthier diets can be prescribed also a personalized ‘online food suggester’ and other cancer preventive measures can be recommended and to those with the symptoms can be advised to go to the doctors with the doctor details.
Current progress on particular problem:
There are many AI based startups to detect this disease at an earlier stage to name a few:
Sascan is one of them which facilitates multispectral camera as a non-intrusive and immediate solution to monitor and detect cancer cells in the mouth. The camera takes pictures of the interior of the mouth with lightings at different wavelengths of light. “The processed images are analysed in real time to decide whether the tissue is abnormal or not,” explains Dr Narayanan.
Aman is another one used to address this challenge by using exosomes, which can be used for molecular diagnosis of cancer without biopsy or scanning. Exocan’s technology-based assay analyses a patient’s biofluids (blood, saliva or urine) to afford exact diagnosis. The results are more precise and cost efficient than usual methods the results are available within just two days. The diagnostic assay, presently under progress, is set to be verified in greater experiments in the upcoming days.
Theranosis is one more startup, which is working on a type of fluid biopsy that detects live cancer cells in blood circulation. Their pioneering “microfluidics lab-on-a-chip” technology takes circulating tumor cells (CTCs) in marginal blood. “The CTCs are live cancer cells that metastasize to other organs via the bloodstream and grow into secondary cancers. So, CTCs are more important than DNA which are derived from dead cancer cells,” Dr Kannan explains.
One more company named Onward is also building a histopathology tool that assists labs procedure twice the amount of samples with the similar availabilities. “One of the tools we are building for a specific set of biomarkers will provide deeper insights and quantified clinical information that will help the oncologist narrow down their choice of treatment decisions which can go a long way in improving patient outcomes,” Dinesh explains. Furthermore, their machine-learning workings for radiology scans mammograms and helps locate suspicious calcifications and masses. It also determines whether these are benign or malignant at an early stage.
Conclusion:
I strongly believe that the proposed methods, can definitely provide an early detection of this deadly disease cancer and will prevent individuals from becoming prey to this disease with the help of enhanced researches, required data sets and experiments. The precision of the results rendered by this model would be more accurate as it is developed with the latest technology ‘Tensorflow - Keras’. This idea can be modeled using Artificial intelligence and machine learning which would serve every human being to overcome the fear of cancer occurrence or curing it at an earlier stage.
References:
[3] A. Amutha and R. S. D. Wahidabanu, "Lung tumor detection and diagnosis in CT scan images," 2013 International Conference on Communication and Signal Processing, Melmaruvathur, 2013, pp. 1108-1112.
[4] K. Gopi and J. Selvakumar, "Lung tumor area recognition and classification using EK-mean clustering and SVM," 2017 International Conference on Nextgen Electronic Technologies: Silicon to Software (ICNETS2), Chennai, 2017, pp. 97-100.
[5] S. Moreno, M. Bonfante, E. Zurek and H. S. Juan, "Study of Medical Image Processing Techniques Applied to Lung Cancer," 2019 14th Iberian Conference on Information Systems and Technologies (CISTI), Coimbra, Portugal, 2019, pp. 1-6.
[6] P. Darshini Velusamy and P. Karandharaj, "Medical image processing schemes for cancer detection: A survey," 2014 International Conference on Green Computing Communication and Electrical Engineering (ICGCCEE), Coimbatore, 2014, pp. 1-6.
[7] T. C. Cahoon, M. A. Sutton and J. C. Bezdek, "Breast cancer detection using image processing techniques," Ninth IEEE International Conference on Fuzzy Systems. FUZZ- IEEE 2000 (Cat. No.00CH37063), San Antonio, TX, USA, 2000, pp. 973-976 vol.2.
[8] W. Wang and S. Wu, "A Study on Lung Cancer Detection by Image Processing," 2006 International Conference on Communications, Circuits and Systems, Guilin, 2006, pp. 371-374.
[9] Y. Lu, J. Li, Y. Su and A. Liu, "A Review of Breast Cancer Detection in Medical Images," 2018 IEEE Visual Communications and Image Processing (VCIP), Taichung, Taiwan, 2018, pp. 1-4.
[10] A. Chaudhary and S. S. Singh, "Lung Cancer Detection on CT Images by Using Image Processing," 2012 International Conference on Computing Sciences, Phagwara, 2012, pp. 142-146.
[11] L. Kapoor and S. Thakur, "A survey on brain tumor detection using image processing techniques," 2017 7th International Conference on Cloud Computing, Data Science & Engineering - Confluence, Noida, 2017, pp. 582-585.
[12] M. Parisa Beham and A. B. Gurulakshmi, "Morphological image processing approach on the detection of tumor and cancer cells," 2012 International Conference on Devices, Circuits and Systems (ICDCS), Coimbatore, 2012, pp. 350-354.
Comentários