Call Girls Kolkata Kalikapur 💯Call Us 🔝 8005736733 🔝 💃 Top Class Call Girl Se...
Artificial intelligence in Caries Diagnosis
1. Deep learning based Neural Network
algorithms in Caries Diagnosis
Presented by
Dr.Mettina
II MDS
Department of conservative dentistry &
Endodontics
2.
3.
4. ABSTRACT
• Objectives: Deep convolutional neural networks (CNNs) are a rapidly emerging new area of medical research,
and have yielded impressive results in diagnosis and prediction in the fields of radiology and pathology. The
aim of the current study was to evaluate the efficacy of deep CNN algorithms for detection and diagnosis of
dental caries on periapical radiographs.
• Materials and methods: A total of 3000 periapical radiographic images were divided into a training and
validation dataset (n = 2400 [80%]) and a test dataset (n = 600 [20%]). A pre-trained GoogLeNet Inception v3
CNN network was used for preprocessing and transfer learning. The diagnostic accuracy, sensitivity, specificity,
positive predictive value, negative predictive value, receiver operating characteristic (ROC) curve, and area
under the curve (AUC) were calculated for detection and diagnostic performance of the deep CNN algorithm.
• Results: The diagnostic accuracies of premolar, molar, and both premolar and molar models were 89.0%
(80.4–93.3), 88.0% (79.2–93.1), and 82.0% (75.5–87.1), respectively.
• Conclusions: This study highlighted the potential utility of deep CNN architecture for the detection and
diagnosis of dental caries. A deep CNN algorithm provided considerably good performance in detecting dental
caries in periapical radiographs.
• Clinical significance: Deep CNN algorithms are expected to be among the most effective and efficient methods
for diagnosing dental caries.
5. INTRODUCTION
• Dental caries are common chronic infectious oral diseases affecting most teenagers &
adults worldwide.
• According to WHO – Dental caries is a microbial multifactorial disease of calcified tissues
of teeth characterized by demineralization of the inorganic portion & destruction of organic
content.
• Classification-
• 1) According to the site of infection
-Pit & fissure caries
-Smooth surface caries
-Root caries
9. • Various restoration methods, for the treatment of dental caries, have been
successfully developed over the past few decades.
• Diagnostic methodology is still a challenge
Deep fissures
Tight
interproximal
contacts
Early stage
Disease
10. • Dental radiography (including panoramic, periapical, and bitewing views) is an
important tool.
• Various diagnostic tools
• Diagnosis is still largely based on empirical evidence.
Visual & tactile Dye tests
FOTI &
DIFOTI
Laser fluorescence
detection
techniques
Alternating current
impedence
spectroscopy
11. ARTIFICIAL INTELLIGENCE
• The term "artificial intelligence" is used to describe machines that mimic "cognitive"
functions that humans associate with other human minds, such as "learning" and
"problem solving".
• The field was founded on the claim that human intelligence "can be so precisely
described that a machine can be made to simulate it".
12.
13.
14.
15. CONVOLUTED NEURAL NETWORKS
• They are an aspect of Artificial intelligence & Deep learning that have demonstrated
excellent performance in computer vision including object, facial & activity recognition,
tracking & three dimensional mapping & localization.
• Medical segmentation and diagnosis is one of the most important fields in which
image processing and pattern recognition procedures have been adopted.
• Diabetic retinopathy, Skin cancer, pulmonary tuberculosis are the areas where Deep
learning based CNN networks have been employed.
16.
17. DEEP LEARNING
• Deep learning (also known as deep structured learning or hierarchical learning) is part
of a broader family of machine learning methods based on artificial neural networks.
Learning can be supervised, semi-supervised or unsupervised.
18. MATERIALS & METHODS
• This study was conducted at the Department of Periodontology, Daejeon Dental Hospital,
Wonkwang University and approved by the institutional Review Board of Daejeon Dental Hospital,
Wonkwang University.
• Anonymized periapicalradiographic image datasets, acquired between January 2016 and
December 2017, were obtained from the authors’ dental hospital’s PACS system (Electronic
medical record)
• All images were clearly revalidated, and dental caries, including enamel and dentinal carious
lesions (excluding deciduous teeth), were distinguished from non-dental caries by four calibrated
board-certified dentists.
19. The dataset consisted of a total of 3000 periapical radiographic images of 778 (25.9%)
maxillary premolars, 769 (25.6%) maxillary molars, 722 (24.1%) mandibular premolars,
and 731 (24.4%) mandibular molars. There were 781 (23.9%) premolars and 772 (25.7%)
molars that were diagnosed as carious, and 719 (26.1%) premolars
and 728 (24.3%) molars diagnosed as non carious.
• Exclusion criteria-moderate-to-severe noise, haziness, distortion and shadows, and those
with a full crown or large partial onlay
20. PRE- PROCESSING & IMAGE AUGMENTATION
• A total of 3000 periapical radiographic images were selected and resized to 299×299
pixels and converted into JPEG file format.
• All maxillary teeth images were reverted to the mandibular teeth form through a
vertical flip. A randomization sequence was created using SPSS (IBM Corporation,
Armonk, NY, USA)
• This was used to divide the dataset into a training and validation dataset (n=2400
[80%]), and a test dataset (n=600 [20%]).
• The training and validation dataset consisted of 1200 dental caries and 1200 non
carious samples and the test dataset consisted of 300 carious and 300 non carious
samples.
21. ARCHITECTURE OF THE CNN
• A pre-trained GoogLeNet Inception v3 CNN network was used for preprocessing, and
the datasets were trained using transfer learning.
• Using -The Inception v3 architecture, which demonstrated excellent performance in the
2014 ImageNet Large Scale Visual Recognition Challenge,
• This has preliminarily learned approximately 1.28 million images consisting of 1000
object categories. It consists of 22 deep layers, and it is possible to obtain different
scale features by applying convolutional filters of different sizes in the same layer.
22. Overall architecture of the pre-trained convolutional neural network model. The
architecture has a total of 9 inception modules, including an auxiliary classifier, and two
fully connected layers. The output layer performs binary classification between dental
caries and non-dental caries using a softmax function
23. STATISTICAL ANALYSIS
The diagnostic accuracy, sensitivity, specificity, positive predictive value (PPV), negative
predictive value (NPV), receiver operating characteristic (ROC) curve, and area under the
ROC curve (AUC) of the test dataset were assessed.
P values<0.05 were considered to be statistically significant, and 95% confidence
intervals (CIs) were calculated
25. Comparison of receiver operating characteristic (ROC) curves for the network
pretrained with augmentation premolar, molar, and both premolar and molar
models. The premolar model had an area under the ROC curve that was significantly greater than
that for molar and both premolar and molar models.
26. DISCUSSION
• Fast and accurate detection and diagnosis are important factors in implementing
appropriate prevention and treatment in patients with dental caries.
• When dental caries are not properly diagnosed, the lesion may progressively invade the
enamel, dentin, and even pulp tissue, inducing severe pain, and ultimately leading to
loss of tooth function.
• By using a deep CNN based on GoogleNet inception V 3 architecture which has
superior performance in image detection classification and segmentation, this study
was able to achieve considerable accuracy & efficiency in detection & diagnosis of
dental caries.
27. • Alongside clinical examinations, radiography plays an important role in the detection of
dental caries, especially for interproximal and root caries.
• Because the settings of various parameters, including brightness, shadow and contrast,
in radiography vary, it is difficult to objectively diagnose dental caries using radiographs
alone.
• Enamel caries can be diagnosed only after the lesion has affected more than one-half of
the enamel thickness.
• Various methods of detection and diagnosis of dental caries that can overcome the
limitations of clinical and radiographic diagnosis are being developed and improved.
28. • Bitewing radiographs and FOTI are useful for the diagnosis of proximal caries. However,
bitewing radiographs have a low diagnostic yield for early dental carious lesions, and
FOTI also yields low diagnostic sensitivity.
• Laser fluorescence has relatively high sensitivity compared with other methods, and can
diagnose early enamel caries.
• Owing to recent advances in digital radiography and computer systems, computer-
assisted diagnostic (CAD) systems are now able to read and interpret radiographs more
precisely. LOGICON Caries Detector™ Software (LCDS), Carestream Dental, GA, USA), ,
was developed to detect and diagnose dental caries based on traditional algorithms
29. • Deep CNN-based dental caries detector can learn location and morphological changes
of dental carious lesions efficiently and detect them conveniently and reliably.
• Deep learning algorithms, such as ResNet and CapsNet, which have deeper or wider
layers, or have modified layering methods are continually being developed as a result,
the accuracy of object detection and segmentation has been significantly and
consistently improved .
• Consider a daily task that we dentists view as routine & relatively simple- finding
caries on X Rays.
Patient
history
Radiographic
exam
Intraoral
examination
Years of
dental
training
30. • For machines to perform tasks such as reading radiographs, they must be trained on
huge data sets to recognize meaningful patterns
• They must understand the spoken language, & be able to make intelligent decisions
regarding that new information & learn from mistakes to improve the decision making
process.
• All this must happen in about the same time that a human being can perform the task.
31.
32. Deep learning for the radiographic detection of apical
lesions Eckert et al
• JOE 2019 May 31..
• Department of Operative and Preventive Dentistry, Charité-Universitätsmediz in Berlin, Berlin, Germany. Electronic
• Introduction: We applied deep convolutional neural networks (CNNs) to detect apical lesions (ALs) on panoramic dental
radiographs.
Methods: Based on a synthesized data set of 2001 tooth segments from panoramic radiographs, a custom-made 7-
layer deep neural network, parameterized by a total number of 4,299,651 weights, was trained and validated via 10
times repeated group shuffling. Hyperparameters were tuned using a grid search. Our reference test was the majority
vote of 6 independent examiners who detected ALs on an ordinal scale (0, no AL; 1, widened periodontal ligament,
uncertain AL; 2, clearly detectable lesion, certain AL). Metrics were the area under the receiver operating characteristic
curve (AUC), sensitivity, specificity, and positive/negative predictive values. .
Results: The mean (standard deviation) tooth level prevalence of both uncertain and certain ALs was 0.16 (0.03) in the
base case. The AUC of the CNN was 0.85 (0.04). Sensitivity and specificity were 0.65 (0.12) and 0.87 (0.04,)
respectively. The resulting positive predictive value was 0.49 (0.10), and the negative predictive value was 0.93 (0.03).
In molars, sensitivity was significantly higher than in other tooth types, whereas specificity was lower.
Conclusions: A moderately deep CNN trained on a limited amount of image data showed satisfying discriminatory
ability to detect ALs on panoramic radiographs.
33. REFERENCE
• Detection & diagnosis of dental caries using a deep learning based
convolutional neural network algorithm Journal of Dentistry 2018
34. •While Science Fiction movies may come to
mind on hearing the word Artificial intelligence,
the future of Artificial intelligence in Dentistry is
very very real……….
Editor's Notes
So we won’t experience 100 years of progress in the 21st century — it will be more like 20,000 years of progress (at today’s rate
In deep learning, a convolutional neural network (CNN, or ConvNet) is a class of deep neural networks, most commonly applied to analyzing visual imagery.
Convolutional networks were inspired by biological processes in that the connectivity pattern between neurons resembles the organization of the animal visual cortex. Individual cortical neurons respond to stimuli only in a restricted region of the visual field known as the receptive field. The receptive fields of different neurons partially overlap such that they cover the entire visual field.
CNNs use relatively little pre-processing compared to other image classification algorithms. This means that the network learns the filters that in traditional algorithms were hand-engineered. This independence from prior knowledge and human effort in feature design is a major advantage.