MADHA INSTITUTE OF ENGINEERING AND

TECHNOLOGY

DETECT SKIN CANCER USING TRANSFORMATIVE

LEARNING
RESNET MODEL
GUIDED BY: PRESENTED BY:
MRS.PARIJATHAM IT.HOD S.SOWNDARYA-211221104015
D.PRIYADHARSHINI-
211221104013
CONTENT:
 ABSTRACT
 INTRODUCTION
 DIAGNOSIS
 CAPSULE NETWORK
 CNN&ANN MODEL
 PROPOSED SYSTEM
 EXISTING SYSTEM
 ABSTRACT:
 Cancer is a deadly disease caused by the uncontrolled proliferation of cells.
 Cancer can occur anywhere in the human body and an involve millions of cells.
 Many people die
from cancer every year, and cancer is considered one of the most serious public health proble
ms.
 Cancer can occur anywhere in the human body and can involve millions of cells.
 Previously, machine learning techniques were used to diagnose skin cancers using a wide vari
ety of proteins and various types of measurements.
 The disadvantage of machine learning methods
is that they have to be done manually, which is very laborious and time-consuming.
 The effectiveness of a single machine learning model in cancer diagnosis is limited.
 INTRODUCTION:
 Skin cancer is the abnormal growth of skin cells most often development of
skin exposed.
 To the sun but the common of cancer can also occur on area of your skin not
ordinarily.
 Exposed to sunlight there are major type of skin cancer based cell carcinoma
and melanoma.
 You can reduce your risk of skin cancer by limiting or avoiding exposure.
 To ultraviolet radiation skin cancer develops primarily on area on sun exposed
skin including the scalp ,face, lips neck chest arms and hand and on areas
that rarely see the light of day you palms beneath your fingernails ore
toenails and you genital area.
DIAGNOSIS:

 A skin cancer is diagnosis involves a physical exam a biopsy and sometime

other test the type depend on the suspected type of skin cancer and extant
of cancer the main test to diagnosis skin cancer.
 to take a sample (biopsy)of the area you need to go to your gp if you are
worried about abnormal area of skin you gp might refer you to a specialist if
they think you have skin cancer or they might do a biopsy themselves if they
have had the specialist training.
CAPSULE NETWORK:

 Capsule network (caps net )are a type of deep nural network that can be
used to classify skin cancer capsnets are designed to improve on the
limitation of convolution neural network (cnns).
 Capsnet can classify skin cancer from dermoscopy image with higher
accuracy the cnns capsnet can detect cancer spot cancer any where in an
image improved caps net such as fix caps have a large receptive filed the
caps net.
CNN&ANN MODEL:
 One of the best method of accurately and sufficient identify skin cancer is using deep
learning in research the deep learning method convolution neural network was used to
direct the who primary malignant and begins using data set.
 The development of computing power the convolutional neural network is on the most
popular malignant and benign development which is one of the deep learning method
provide high success in basic analysis.
 Artificial neural network are used skin cancer detection system to image as melanoma
or non melanoma ANN
 A machine learning algorithms that uses a network of artificial neurons to solve.
 ANN are inspired by the structure of human brain neurons that send electrical signal to
each other.
CNN ARCHITECTURE:
EXISTING SYSTEM:
 one of the best of accurately and sufficient skin cancer using deep
learning (dl)the research of deep learning method convolution neural
network (CNN)was used to detect the two primary type of tumors
malignant and benign using ISIC data set.
 a dermatologist may ask you if you noticed change in any existing moles
or other skin sport or if you notices any new skin growth.
 skin cancer the abnormal growth of skin cells most after develop on skin
exposed to the sun but the common form of cancer also occur on of your
skin not ordinary exposed sunlight there are three major type of skin
cancer basal cell carcinoma squamous cell carcinoma and melanoma.
PROPOSED SYSTEM:

 A method based on artificial neural network for cancer diagnosis the method
used asymmetry border color diameter for cancer analysis result should
about 97% accuracy for the training of neural network the method is then
used for testing.
 The cancer classification skin cancer proposed a four stage skin cancer
detection method preprocessing segmentation feature extraction and
classification this method using CNN convolutional neural network.
ARCHITECTURE:
i
[Feature
Extraction(ResNet)]
[Input Layer] [Preprocessing]

[Convolutional Layers]

[Image Resizing]
[Normalization]
[Data Augmentation] [Fully Connected Layers]
[Data Augmentation] [Pooling layer]

[Output Layer]

[Loss Function]

[Optimization

[Model Training]

[Evaluation Metrics]

[Prediction (Inference)]
Modules:
1. Input Layer
2. Preprocessing
3. Pooling Layer
4. Output Layer
LITERATURE SURVEY

Literature survey:
Tittle: deep residual network for skin cancer deduction
Author name: robert johnson
Abstract:
 Skin cancer is one of the most common forms of cancer globally with
melanoma being the deadlist
 The literature survey review various research efforts utilize resnet based
model skin cancer deduction.
Authers :john doe jane smith
Apstract:
Tittle :skin cancer deduction
Skin cancer is the one of the form of cancer world wide with early
detection using a crucial role successful treatment
In literature survey using detection skin cancer focasin primary on
image processing
Tittle: advance skin cancer
Auther name: john smith
Author name:
 Skin cancer remain one of the most prevalent form of cancer
 Advance technology using in method skin cancer deep learning
 The literature survey innovation in automation skin cancer deduction.
TITTLE: Skin cancer deduction using resnet model
ABSTRACT:
 Critical filed a medical diagnostics and research advertisement have focused
on appling resent model to improve detection
 The review of various models used in the skin cancer computional model and
real world.
CONCLUSION:
In conclusion, using a transformative learning-based resnet model
for skin cancer detection is a promising approach in medical image
analysis.
The resnet architecture, known for its ability to learn complex
representations through residual learning, helps in addressing
vanishing gradient issues, enabling deeper networks to be trained
efficiently.
When combined with transformative learning techniques, which
focus on enhancing the model’s ability to generalize and adapt
through iterative refinement, it can significantly improve the
accuracy and robustness of skin cancer detection systems.
REFERENCES

(1) A. Afhoa, R. Z. A. O. Gania, M. U. Khan, U. Jilam, And K. M. Ahmed, “Skin Lenon

Classification Using Machine Learning Approach: A Survey 2022 Global Conference On
Wireless And Optical Technologies”

(2)N.C. Tp//Www.Ca.Go Foundation, M., Https://Wanoma.Org/.

(3) 2. Wide A. Waheed. M. Zafas, And F. Rius, “An Efficiens Machine Loaming Approach The
Detection Of Melanoma Using Dermoscopic Images In 2017 Isrnational Conference On
Communication, Computing And Digital Systems (C-CODE), 2017, Pp. 316-319

(4) M. Saltar And A. Magd. “Lang Cancer Classification Models Ng Discrim Namtan Of
Mutated Genes In Guten Amino Acids Sunces An Huan Journal For Science And Engmering,
Vol. 44, 2019 4, Pp. 3197-3211.
(5) A. Maid. S. Al M. Igbul, And N. Kassat, Prediction Of Human Becait And Onlon Cancers
From Imbalanced Data Using Ared Weight And Suppart Vector Machines, Conquer Methods
And Programs In Hiomedicine, Vol. 113, 00. 3, Pp. 742-80k, 2014.
(6) R. Seega And A. Suresh, “Deep Lesening Based Skin Lesion Segmentation And [ Classification
Of Melanoma Ating Support Vector Machine (SVM) A Pacific Pournal Of Cancer Prevention APICP,
Vol. 20, 5. P. 1555, 2010.
(7) Y. Yaan, M. Chand Y-C Lo “Automatic Skan Less Segmentation Uung Deep Fully Convaltional
Netwerks With Jaccard Distance, IEEE Transactions On Medical Asaging, Vol. 36, A. 9, Pp. 1876-
1886 2017
(8) L. Yu. Ii. Chen, Q. Dest. 1. Qin, And P-A. Hong. “Automated Netamoma Seongsition In
Dermescopy Iniages Via Very Deep Residual Network,” IEEE Transactions On Medical Imaging, Vol.
36, 40. 4, Pp. 994-1004, 2016
(9) 1 Bi. 1. Kim, E Aha, A A. Kumar, M. Fulham, And D. Feng, “Dermoscopic Image Segmentation
Mahistage Fully Cursulutumal Networks, IEEE Trasactions On Biomedical Engineering, Vol. 64, No.
9, Pp. 2065-2074, 2017
[10] U-0 Dorj, K-K. Lee, L-Y. Class, And M. Len. “The Skin Cancer Classi Fication Using Dorp
Convolutimal Arural Network. Mutimedia Yodis Andmapplications, Vol. 77, Пл. 8, Pp. 939-9924.
2018.