40
AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1 , R. Leonardi 2 , F. Maiorana 1 , G. Cristaldi 1 , M.L. Distefano 2 1 Dipartimento di Ingegneria Informatica 2 Clinica Odontoiatrica II - Policlinico University of Catania Italy

AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Embed Size (px)

Citation preview

Page 1: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

AUTOMATIC LANDMARKING OF CEPHALOGRAMS

BY CELLULAR NEURAL NETWORKS

D. Giordano1, R. Leonardi2, F. Maiorana1, G. Cristaldi1, M.L.

Distefano2 1Dipartimento di Ingegneria Informatica

2Clinica Odontoiatrica II - Policlinico

University of CataniaItaly

Page 2: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Cephalometric analysis

• Cephalograms are lateral skull radiographs taken under standard conditions

• Cephalometric analysis is based on the identification of landmarks, which are used for linear and angular measurements

• It is important for orthodontic planning and treatment evaluation

Literature review Outlines of CNNs. Tool and the CNN

templates Experimental evaluationResultsConclusions

Page 3: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

A cephalogram

Page 4: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Tracing key anatomical structures

Page 5: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Landmarks identification

Baseline for measurements

Page 6: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Approaches to cephalometrics

1. Manual. placing a sheet of acetate over the cephalometric radiograph, tracing salient features, identifying landmarks and measuring distances and angles between landmark location.

2. Computer aided. Landmarks are located manually while these locations are digitized into a computer system. The computer then completes the cephalometric analysis.

3. Completely automated. The computer automatically locates landmarks and performs the cephalometric analysis.

AIME 05

Literature review Outlines of CNNs. Tool and the CNN

templates Experimental evaluationResultsConclusions

Page 7: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

1. speed-up a very time-consuming manual process

2. improve measurements accuracy

AIME 05

Why automated landmarking?

Literature review Outlines of CNNs. Tool and the CNN

templates Experimental evaluationResultsConclusions

Page 8: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

PRIOR KNOWLEDGE LEARNING APPROACH

AIME 05

Previous approaches to automated landmarking

Literature review Outlines of CNNs. Tool and the CNN

templates Experimental evaluationResultsConclusions

Page 9: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

1. Use of filters to minimize noise and enhance the image,

2. Application of operators for edge detection,

3. On line-following algorithms guided by a prior knowledge, introduced in the system by means of simple ad hoc criteria

AIME 05

Approaches based on prior knowledge

Literature review Outlines of CNNs. Tool and the CNN

templates Experimental evaluationResultsConclusions

Page 10: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Some examples of the techniques that have been used:

• Neural networks together with genetic algorithms

• Fuzzy neural networks• Active shape models

AIME 05

Approaches based on learning and pattern matching

Literature review Outlines of CNNs. Tool and the CNN

templates Experimental evaluationResultsConclusions

Page 11: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Work Sample size

Techniques

Parthasarathy et al.

(1989)

5 Resolution piramidKnowledge based line

extractor

Tong et al. (1990)

5 Resolution pyramid Edge enhancementKnowledge-based extraction

Cardillo et al.(1994)

40 Pattern matching

Rudolph et al.(1998)

14 Spatial spectroscopy Statistical pattern

recognition

Liu et al.(1999)

38 Multilayer Perceptron Genetic Algorithms

Hutton et al.(2000 )

63 Active Shape Models

El-Feghi et al.(2003)

200 Fuzzy neural network

Innes et al.(2002)

109 PCNN pulse coupled neural networks

Page 12: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Limitations of previous approaches

1. Accuracy achieved

2. Performance varying on different landmarks

3. Strongly dependent on the quality of the X-rays

Golden standard: landmarks should be located within 1mm tolerance; although 2mm is deemed acceptable for clinical practice

AIME 05

Literature review Outlines of CNNs. Tool and the CNN

templates Experimental evaluationResultsConclusions

Page 13: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Our approach

• The proposed method proposed is based on CNN (Cellular Neural Networks)

• CNNs are an emerging paradigm for image processing

• CNNs is a powerful computational model equivalent to a Turing Machine

AIME 05

Literature review

Outline of CNNsTool and the CNN

templates Experimental evaluationResultsConclusions

Page 14: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Cellular Neural Networks

• CNN consist of computational units (cells) arranged in matrix forms (2D) or cube forms (3D)

• Each cell is a dynamic unit with an input, an output and a state

• Each cell is influenced by the input and the output of all neighboring cells within a given radius

AIME 05

Literature review

Outline of CNNsTool and the CNN

templates Experimental evaluationResultsConclusions

Page 15: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

The neighborhood circle of the interacting cells is defined as follows:

Nr (i,j) = C (k,h): max ( k-i , h-j ) ≤ r, 1≤ k≤ M; 1≤ h≤ N

where M and N are the matrix dimensions

AIME 05

Circles of influence with radius equal to one for cells

Cij, Ci+1, j+1

Literature review

Outline of CNNsTool and the CNN

templates Experimental evaluationResultsConclusions

Cellular Neural Networks

Page 16: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

• CNN dynamics are determined by the following equation, where x is the state, y is the output, u is the input,

• xij is the generic cell belonging to the matrix

• Iij is the activating treshhold for each cell.

AIME 05

ItuhkjiBtyhkjiAxxjiNhkC

khjiNhkC

khjiji

rr

)(),,,()(),,,(),(),(),(),(

,

.

,

)11(2

1,,, jijiji xxy

Cellular Neural Networks

Literature review

Outline of CNNsTool and the CNN

templates Experimental evaluationResultsConclusions

Page 17: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

• CNN dynamics are determined by the following equation, where x is the state, y is the output, u is the input,:

• A is known as feedback template

• B is known as control template

AIME 05

ItuhkjiBtyhkjiAxxjiNhkC

khjiNhkC

khjiji

rr

)(),,,()(),,,(),(),(),(),(

,

.

,

)11(2

1,,, jijiji xxy

Cellular Neural Networks

Literature review

Outline of CNNsTool and the CNN

templates Experimental evaluationResultsConclusions

Page 18: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

• Several image processing tasks can be performed by CNNs by programming by templates

• Library of known templates are available

• A key advantage is that the inherently parallel architecture of the CNN can be implemented on chips, known as CNN-UM (CNN Universal Machine) chips allowing computation times three orders of magnitude faster than classical methods.

Cellular Neural Networks

Literature review

Outline of CNNsTool and the CNN

templates Experimental evaluationResultsConclusions

Page 19: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

In our work we used:• A constant treshold for each cell

• A circle of influence with radius equal to 1 (A, B: 3X3) and with radius equal to 2 (A, B: 5X5)

• Every cell has an initial state variable equal to zero

• Contour condition uij = 0 (Dirichlet condition)

• Input: the image to be processed

• Symmetrical feedback templates (to ensure steady state)

• Exploitation of the transient solution n. of cycles and integration steps are important for landmark identification

AIME 05

Cellular Neural Networks

Literature review

Outline of CNNsTool and the CNN

templates Experimental evaluationResultsConclusions

Page 20: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Our system is based on a software simulator of a CNN of 512X480 cells.

It uses different types of CNNs on the scanned cephalogram

1) first to pre-process the image and eliminate the noise,

2) then to ensure that each landmark region is properly highlighted (by appropriate CNN templates)

3) landmark-specific algorithms including expert rules for point identification are then applied and landmarks coordinates computed

AIME 05

Tool Literature review Outline of CNNs

Tool and CNN templates

Experimental evaluationResultsConclusions

Page 21: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

AIME 05

Page 22: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

The system operates based on two classes of rules

• Expert rules concerning

where landmark should be located,

• Rules to select the proper CNN template based on local image properties

AIME 05

Tool Literature review Outline of CNNs

Tool and CNN templates

Experimental evaluationResultsConclusions

Page 23: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

The tool has been designed to detect 8 landmarks, which are essential to conduct a basic cephalometric analysis:

• Menton, • B point, • Pogonion, • PM point, • A point, • Upper incisal, • Lower incisal, • Nasion.

AIME 05

Tool Literature review Outline of CNNs

Tool and CNN templates

Experimental evaluationResultsConclusions

Page 24: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Why n. of cycles are important

AIME 05

Non saturated CNN Output Saturated CNN Output

Using images with the same brightness simplifies point extraction and emphasize program correctness

Literature review Outline of CNNs

Tool and CNN templates

Experimental evaluationResultsConclusions

Page 25: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Menton

AIME 05

;

00000

00000

00100

00000

00000

A ;

11111

00000

00000

00000

11111

B

Templates and Templates and CNN output for CNN output for Menton Menton (n.cycles=30) (n.cycles=30)

Literature review Outline of CNNs

Tool and CNN templates

Experimental evaluationResultsConclusions

Page 26: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Gnation and B point

Templates and CNN Templates and CNN output for Chin output for Chin Curvature Curvature (n.cycles=30) (n.cycles=30)

;

000

010

000

A ;

110

101

011

B

Literature review Outline of CNNs

Tool and CNN templates

Experimental evaluationResultsConclusions

Page 27: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

AIME 05

Up and low Incisors

Templates and CNN output for Templates and CNN output for incisors Curvature (n.cycles=60)incisors Curvature (n.cycles=60)

00000

10001

10001

00000

00000

B

00000

20002

20002

00000

00000

B

Good contrast Good contrast and luminosityand luminosity

Low contrast Low contrast and luminosityand luminosity

Literature review Outline of CNNs

Tool and CNN templates

Experimental evaluationResultsConclusions

Page 28: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Nasion

White nasion Black nasion

Four templates were used

Literature review Outline of CNNs

Tool and CNN templates

Experimental evaluationResultsConclusions

Page 29: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento
Page 30: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

AIME 05

Page 31: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

• 8 landmarks were chosen for preliminary assessment of the method, and a set of 97 digital X-rays was landmarked by an expert orthodontist.

Literature review Outline of CNNsTool and CNN templates

Experimental evaluation

ResultsConclusions

Assessment

Page 32: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

• The first stage assessed the image output of the CNNs, to verify that it included the sought landmark.

• This was done by visual inspection from the same expert who landmarked the X-rays.

• Over 97 cases, 29 cases (30%) led to CNN outputs in which some edges were overly eroded. This implies that the number of processing cycles in these cases needs to be reduced.

AssessmentLiterature review Outline of CNNsTool and CNN templates

Experimental evaluation

ResultsConclusions

Page 33: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

• The second stage evaluated performance of the developed algorithms for 8 landmarks

• Sample of 26 cases randomly selected from the previous one after eliminating the cases that had not been taken into consideration by the algorithms.

AssessmentLiterature review Outline of CNNsTool and CNN templates

Experimental evaluation

ResultsConclusions

Page 34: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

The coordinates of each point found by the program were compared to expert landmarking, and the Euclidean distance of the found landmark from the reference

one was computed.

AssessmentLiterature review Outline of CNNsTool and CNN templates

Experimental evaluation

ResultsConclusions

Page 35: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Results

Landmark Mean error(mm)

MD SD ≤1 (mm)

>1;≤2(mm)

Imprecise cases Success Rate

Success Rate

(overall sample)

≤3 (mm)

>3 (mm)

Upper incisor

.48 .25 .60 88% 8% 4% - 96% 92%

Lower incisor

.92 .67 .94 66% 26% 4% 4% 92% 81%

Nasion 1.12 .76 1.11 70% 17% - 13% 87% 81%

A Point 1.34 1.06 .82 58% 21% 17% 4% 79% 73%

Menton .62 .33 .82 85% 7% 4% 4% 92% 92%

B Point 2.00 .42 3.3 71% 8% - 21% 79% 73%

Pogonion .87 .04 1.34 73% 8% 8% 11% 81% 81%

PM Point 1.25 .33 1.68 69% 8% 8% 15% 77% 77%

Literature review Outline of CNNsTool and CNN templates Experimental evaluation

ResultsConclusions

Page 36: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Work and Ref. Sample size

N. Landmarks and accuracy Techniques

Parthasarathy et al. (1989) [10]

5 9 landmarks, 58% < 2mm, (18%<1mm)

mean error: 2.06 mm

Resolution piramidKnowledge based line extractor

Tong et al. (1990) [11]

5 17 landmarks, 76%< 2mmmean error: 1.33 mm

Resolution pyramid Edge enhancementKnowledge-based extraction

Cardillo et al.(1994) [13]

40 20 landmarks, 75% < 2mm mean error: not reported

Pattern matching

Rudolph et al.(1998) [14]

14 15 landmarks, 13% <2mm mean error: 3,07 mm

Spatial spectroscopy Statistical pattern recognition

Liu et al.(1999) [6]

38 13 landmarks, 23% < 2mm(8% <1mm),mean error: 2,86 mm

Multilayer Perceptron Genetic Algorithms

Hutton et al.(2000 ) [7]

63 16 landmarks, 35% < 2mm(13% < 1mm)mean error: 4,08

Active Shape Models

El-Feghi et al.(2003) [16]

200 20 landmarks, 90% <2mmmean error: not reported

Fuzzy neural network

Innes et al.(2002) [18]

109 3 landmarks, 72% <2mm, mean error: not reported

PCNN : pulse coupled neural networks

Our Work 26 8 landmarks, 85%<2mm (73% < 1mm)

mean error: 1.07 mm

Cellular Neural NetworksKnowledge based landmark

extraction

Page 37: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

The experimental results have shown that of the

sought landmarks 85% are within 2mm

precision, and remarkabily that 73% are

within 1mm.

ResultsLiterature review Outline of CNNsTool and CNN templates Experimental evaluation

ResultsConclusions

Page 38: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

Conclusions

• CNNs provide an effective method to pre-process images for automated landmarking

• They are accurate and flexible (integration of edge based and region based methods)

• Their hardware implementation affords real-time performance

Literature review Outlines of CNNs. CNN prototype and the

templates Reports the experimental

evaluationResultsConclusions

Page 39: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

The approach that we have employed will be further improved by prior classification on the cases based on:

1. Key morphologies of the skull (e.g., byte typology, shape of anatomical structures)

2. X-ray brightness

ConclusionsLiterature review Outline of CNNsTool and CNN templates Experimental evaluationResults

Conclusions

Page 40: AUTOMATIC LANDMARKING OF CEPHALOGRAMS BY CELLULAR NEURAL NETWORKS D. Giordano 1, R. Leonardi 2, F. Maiorana 1, G. Cristaldi 1, M.L. Distefano 2 1 Dipartimento

MANY THANKS

GrazieGrazie