Automated Carotid IMT Measurement and Its Validation in Low Contrast Ultrasound Database of 885 Patient Indian Population Epidemiological Study: Results of AtheroEdge^® Software

Abstract

This chapter demonstrates the usage of an automated computer-based IMT measurement system called CALEX 3.0 (a class of patented AtheroEdge^® software) on a low contrast and low-resolution image database acquired during an epidemiological study from India. The overall purpose of this chapter is to show that fully automated methods nowadays have accuracy and reproducibility suitable to epidemiological studies.

The image contrast was very low resolution, with pixel density of 12.7 pixels/mm. The accuracy and reproducibility of the AtheroEdge^® software system were compared with that of the manual tracings of a vascular surgeon—considered as a gold standard.

We automatically measured the IMT value of 885 common carotid arteries in longitudinal B-Mode images. CALEX 3.0 consisted of a stage for the automatic recognition of the carotid artery and an IMT measurement modulus made of a fuzzy K-means classifier. Performance was assessed by measuring the system accuracy and reproducibility against manual tracings by experts.

Results were very encouraging: CALEX 3.0 processed all the 885 images of the dataset (100% success). The average automated obtained IMT measurement by CALEX 3.0 was 0.407 ± 0.083 mm compared with 0.429 ± 0.052 mm for the manual tracings, which led to an IMT bias of 0.022 ± 0.081 mm. The IMT measurement accuracy (0.022 mm) was comparable to that obtained on high-resolution images and the reproducibility (0.081 mm) was very low and suitable to clinical application. The Figure-of-Merit defined as the percent agreement between the computer-estimated IMT and manually measured IMT for CALEX 3.0 was 94.7%.

CALEX 3.0 had a 100% success in processing low contrast/low-resolution images. CALEX 3.0 is the first technique, which has led to high accuracy and reproducibility on low-resolution images acquired during an epidemiological study. We propose CALEX 3.0 as a generalized framework for IMT measurement on large datasets.

Appendices

Appendix 1. Polyline Distance

The polyline distance metric (PDM) is a robust metric to define the distance between two boundaries. The basic idea is to measure the distance of each vertex of a boundary to the segments of the other boundary. The polyline distance from vertex v to the boundary B ₂ can be defined as the minimum distance between v and the segments of B ₂. The distance between the vertexes of B ₁ and the segments of B ₂ is then defined as the sum of the distances from the vertexes of B ₁ to the closest segment of B ₂. Let’s call this distance as d(B ₁,B ₂). Similarly, it is possible to calculate the distance between the vertices of B ₂ and the closest segment of B ₁ (let’s call this distance as d(B ₂,B ₁)). The polyline distance between boundaries is the defined as

$$ D\left( {{B_1},{B_2}} \right)=\frac{{d\left( {{B_1},{B_2}} \right)+d\left( {{B_2},{B_1}} \right)}}{{\left( {\#\ \mathrm{of}\ \mathrm{vertices}\ \mathrm{of}\ {B_1} + \#\ \mathrm{of}\ \mathrm{vertices}\ \mathrm{of}\ {B_2}} \right)}} $$

(17.1)

Appendix 2. Definition of the IMT Bias, Absolute Error, and Squared Errors

Let IMT _i be the intima–media thickness value automatically computed by CALEX 3.0 on the ith image of the database. Let GTIMT _i be the IMT value computed by manual measurements.

The IMT measurement bias ε _i is defined as

$$ {\varepsilon_i}=\mathrm{GTIM}{{\mathrm{T}}_i}-\mathrm{IM}{{\mathrm{T}}_i} $$

(17.2)

The absolute value μ _i of the IMT bias is defined as

$$ {\mu_i}=\left| {\mathrm{IM}{{\mathrm{T}}_i}-\mathrm{GTIM}{{\mathrm{T}}_i}} \right| $$

(17.3)

The squared error η _i is, finally, defined as

$$ {\eta_i}={{\left| {\mathrm{IM}{{\mathrm{T}}_i}-\mathrm{GTIM}{{\mathrm{T}}_i}} \right|}^2} $$

(17.4)

By averaging all these error metrics on the N images of the database, we computed the overall system errors as:

$$ \overline{\varepsilon}=\frac{1}{N}\sum\limits_{i=1}^N {{\varepsilon_i}} $$

(17.5)

$$ \overline{\mu}=\frac{1}{N}\sum\limits_{i=1}^N {{\mu_i}} $$

(17.6)

$$ \overline{\eta}=\frac{1}{N}\sum\limits_{i=1}^N {{\eta_i}} $$

(17.7)

Appendix 3. Figure-of-Merit

Let IMT _i be the intima–media thickness value automatically computed by CALEX 3.0 on the ith image of the database. Let GTIMT _i be the IMT value computed by manual measurements. If we consider a database of N images, then the overall system IMT estimate can be defined as

$$ \overline{\mathrm{IM}\mathrm{T}}=\frac{1}{N}\sum\limits_{i=1}^N {\mathrm{IM}{{\mathrm{T}}_i}}$$

(17.8)

$$ \overline{\mathrm{GTIM}\mathrm{T}}=\frac{1}{N}\sum\limits_{i=1}^N {\mathrm{GTIM}{{\mathrm{T}}_i}} $$

(17.9)

The Figure-of-Merit (FoM) is mathematically represented as

$$ \mathrm{FoM}=100-\left| {\frac{{\overline{\mathrm{IMT}}-\overline{\mathrm{GTIMT}}}}{{\overline{\mathrm{GTIMT}}}}} \right|\times 100 $$

(17.10)