Beruflich Dokumente
Kultur Dokumente
Shreyas Akella
UWM ID 991360075
1
Abstract
Cancer is a dangerous disease with evidence based discussions, suggestive of
long-term effects. The evidence is usually generated from exterior of an organism,
and it usually comes from extensive medical quantitative analysis. A few examples
covered here will range from skin cancer to colon cancer. A number of trends are
noted, including a growing dependence on protein biomarkers and microarray
data, a strong bias towards applications in prostate and breast cancer, and a heavy
reliance on “older” technologies such artificial neural networks (ANNs) instead of
more recently developed or more easily interpretable machine learning
methodsThis paper aims to discuss the areas where Machine Learning(ML) and
Neural Networks(NN) and applicable and what is the result of doing so. This paper
will also discuss the individual progress treatments, and the difference between
incorporation of AI and the usual techniques used. In assembling this review, I’ve
surveyed many papers (referenced below), while drawing conclusions from a few.
Introduction
Cancer is a very degenerative disease, and the prediction of this disease has been
quite challenging for decades, until machine learning methods have been applied
to cancer detection. The fundamental goals of cancer prediction and prognosis are
distinct from the goals of cancer detection and diagnosis. In cancer
prediction/prognosis one is concerned with three predictive foci: the prediction of
cancer susceptibility (i.e. risk assessment); the prediction of cancer recurrence and
the prediction of cancer survivability.
There are a lot of factors, when it comes to detecting cancer. Most of the details
usually lie with the patient, and they include height, weight, history, allergies,
treatments for said allergies, smoking behaviour (for lung cancer), age, location and
environmental upbringing. Other issues that do affect prognosis, but are out of the
range of a patient’s control are the location of the cancer, whether it has
metastasized, whether it is degenerative, etc. Usually, these details require very
2
specific molecular details, about the patient’s genetic make-up or the tumor
present.
With the advent of new techniques like genome sequencing, and with the
improvement of other techniques like tissue arraying, medical imaging, etc, these
molecular details have been very easy to obtain.
Artificial neural networks (ANNs) are regression devices containing layers of
computing nodes (crudely analogous to the mammalian biological neurons) with
remarkable information processing characteristics. They can detect sways from
linear regressions that are not explicitly formulated as inputs, making them capable
of learning and adaptability. They possess high parallelism, robustness,
generalization and noise tolerance, which make them capable of clustering,
function approximation, forecasting and association, and performing massively
parallel multifactorial analyses for modeling complex patterns, where there is little
prior knowledge.
Specifications
In this section, I have reviewed three different types of cancer, namely, colon
cancer, breast cancer and skin cancer. The papers I’ve reviewed talk extensively
about employing ANN ‘s and CNN’s (convoluted neural networks), in a way that
addresses the prognostic definitions. Each of these cancers have a different
approach towards artificial intelligence.
COLON CANCER
In this scenario, Artificial neural networks (ANNs) have been employed. They are
rigorously adaptive and help the database capacity tenfold. They possess
non-linearity, noise insensitivity and high parallelism. This helps to gain a better
understanding of how the cancer advances, as a pattern, and helps identify the
nature and any potential risks evaluated.
3
When an ANN is given a particular set of data,it recognizes patterns in the data, and
builds an analogous system to predict the same outcomes for that particular data
set . Training, which is a synonym to biological learning, is carried by a "teacher"
program that loads in training cases from a database and adjusts the weights and
thresholds value of the network to minimize the error between the real-world
outputs and the network generated outputs for the training case inputs.
The benefit of an ANN is that it can learn from a multitude of variable analyses.
Although larger data sets mean a bigger deviation from the actual probabilistic
output, the neural network usually provides a method to minimize errors. This
helps in the expansion of local knowledge to a more global database.
As of now, it is quite tough to predict the advancement of colon cancer and its
variants on the chemotherapeutic release. It is also difficult to predict the death of
said patient with colon cancer at a quite advanced stage (B or C).
In the particular paper I’ve studied, an ANN was trained from a 5 year follow up
data set that included 334 patients with colorectal cancer. This ANN was trained on
these data sets and another follow up for 12 months. This was to validate around
50 patients and to predict death within 9, 12, 15, 18 and 21 months. This ANN was
90% accurate when it came to predicting deaths. It had an overall accuracy of 79%
when compared with clinicopathological methods.
In the other study, in the UK, an ANN was used to predict the survival rate of
patients suffering from colon carcinoma. An FFNN with 2 hidden layers with 4 and 3
hidden neurons, respectively, was selected. The ANN yielded a ROC area of 87.6%.
At sensitivity to mortality of 95%, the specificity was 41%. The logistic regression
yielded a ROC area of 82%, and sensitivity to mortality of 95% gave a specificity of
only 27%.
SKIN CANCER
4
Esteva et al, provided 129,450 different images of 2032 different skin lesions. Out of
these, 3374 images were obtained from a dermatoscopic device. The trained CNN
was tested with test data that were fully biopsy-proofed and achieved an ROC AUC
of .96 for carcinomas, an ROC AUC of .96 for melanomas, and an ROC AUC of .94
for melanomas classified exclusively with dermatoscopic images.
BREAST CANCER
Breast cancer is a survivable cancer that, if caught early, is treatable with a very
high success rate. Unfortunately, early detection is quite a challenging process, and
with the exception of self diagnosis, there was no way to tell if a patient has
advanced cancer. If detected early the chance of survival goes from a mere 56% to
a very high 85%.
In the paper I’ve reviewed on breast cancer and its classifications, the proposed way
to detection was the rule-based classification, for the prediction of survival.
The data set used 8 attributes. This included a record of 900 patients, with 876
being female, and 24 patients being male.
Naive Bayes (NB), Trees Random Forest (TRF), 1-Nearest Neighbor (1NN), AdaBoost
(AD), Support Vector Machine (SVM), RBF Network (RBFN), and Multilayer
Perceptron (MLP) machine learning techniques with 10-cross fold technique were
used with the proposed model for the prediction of breast cancer survival. The
performance of machine learning techniques were evaluated with accuracy,
precision, sensitivity, specificity, and area under ROC curve.
Out of these patients, 93, unfortunately died, and 807 patients were alive. In this
study, Trees Random Forest (TRF) technique showed better results in comparison
to other techniques. The accuracy, sensitivity and the area under the ROC curve of
TRF are 96%, 96%, 93%, respectively.
6
This study demonstrated that Trees Random Forest model (TRF) which is a
rule-based classification model was the best model with the highest level of
accuracy. Therefore, this model is recommended as a useful tool for breast cancer
survival prediction as well as medical decision making.
I. SKIN CANCER
The first issue with skin lesions was that the considered formulations of the
images only slightly different. This wasn’t the case just with the training data,
but the used data and the output as well. This meant a reduced accuracy,
and reduced reliability on such incremental methods.
Another challenge was the updating of the image directory. The lesions that
were used as a part of the project only included images of light coloured
people. This produced errors for people of different colour. An updated
global image registry means a representation of the world population.
There exist data barriers beyond which mathematical models are unable to
make predictions in biological systems.
Regression models are superior to ANNs when drawing inferences and
interpretations based on outputs.
CONCLUSIONS
The incentive, to improve cancer detection through neural networks isPlausible,
based on the positive neural adjustments. Systematic optimization of each neural
network that differ in their approach to optimization contribute to highly predictive
network. The definitive high level of classification accuracy, sensitivity, and
specificity indicated by the results has the potential to dramatically increase in
Clinical management. Specifically the technology could provide accurate prognostic
Value, as a precursor to histologic or cytologic evaluation of biopsied tissue helping
on a pathological level.
Unfortunately, it is difficult and in many cases, impossible to compare the
performance of published classification results since many authors use nonpublic
datasets for training and/or testing.
While ANNs still predominate it is evident that a growing variety of alternate
machine learning strategies are being used and that they are being applied to many
types of cancers to predict at least three different kinds of outcomes. It is also clear
that machine learning methods generally improve the performance or predictive
accuracy of most prognoses, especially when compared to conventional statistical
or expert-based systems.
REFERENCES
1. Skin Cancer Classification Using Convolutional Neural Networks: Brinker et al
2. Neural Networks and Other Machine Learning Methods in Cancer Research -
Alfred Vellido, et al. - University of Byrom St.
8
3. Efficient Cancer Detection Using Multiple Neural Networks - IEEE Health and
science Journal. John Shell, et al.
4. Applications of Machine Learning in Cancer Prediction and Prognosis - Joseph
A. Cruz, David S. Wishart - University of Alberta Edmonton - John A Cruz et al.
5. Machine Learning models in Breast cancer survival predictions - Medical
Informatics Research Center, Montezari.M et al.
6. Artificial neural networks for diagnosis and survival prediction in colon
cancer - Farid Ahmed - Dept of oncology, Brody School of Medicine