Classification of hysteroscopical images using texture and vessel descriptors

Abstract

In recent years, hysteroscopy, used as an outpatient office procedure, in combination with endometrial biopsy, has demonstrated its great potential as the method of first choice in the diagnosis of various gynecological abnormalities including abnormal uterine bleeding (AUB) and endometrial cancer (CA). In patients suffering with AUB, the blood vessels of the endometrium are hypertrophic, whereas in the case of CA vascularization is irregular or anarchic. In this paper, a methodology for the classification of hysteroscopical images of endometrium using vessel and texture features is presented. A total of 28 patients with abnormal uterine bleeding, 10 patients with endometrial cancer and 39 subjects with no pathological condition were imaged. 16 of the patients with AUB were premenopausal and 12 postmenopausal, all with CA were postmenopausal, and all with no pathological condition were premenopausal. All images were examined for the appearance of endometrial vessels and non-vascular structures. For each image, 167 texture and vessel’s features were initially extracted, which were reduced after feature selection in only 4 features. The images were classified into three categories using artificial neural networks and the reported classification accuracy was 91.2 %, while the specificity and sensitivity were 83.8 and 93.6 % respectively.

Appendix

1.1 The sequential floating forward search (SFFS) method

Suppose $D$ is the number of the available features ($D = 167$), $i(1 \le i \le D)$ is a feature and $X_{k}$ is a subset of $k$ features. The SFFS method requires the existence of a function $J$ that evaluates the “goodness” of the feature subset under consideration.

• Step 1: $X_{0} = \emptyset$ and $k = 0$

• Step 2: While $k < 2$

  • Find the most significant feature $y = \mathop {\arg \hbox{Max} }\limits_{{a \in (Y - X_{k} )}} J(X_{k} \cup \{ a\} )$

  $$ X_{k + 1} = X_{k} \cup y $$
  $$ k = k + 1 $$

• Step 3: While $k < D$

  • Find the most significant feature $y = \mathop {\arg \hbox{Max} }\limits_{{a \in (Y - X_{k} )}} J(X_{k} \cup \{ a\} )$

  $$ X_{k + 1} = X_{k} \cup y $$

  • While $k < 2$

    Find the least significant feature $x = \mathop {\arg \hbox{Max} }\limits_{{a \in X_{k} }} J(X_{k} \cup \{ a\} )$

    If $J(X_{k} - \{ x\} ) > J(X_{k - 1} )$

    $$ X_{k - 1} = X_{k} - \{ x\} $$
    $$ k = k - 1 $$

The algorithm is initialized by setting $X_{0} = \emptyset$ and $k = 0$ and the feature selection technique is used until the set $X_{k}$ includes the $k$ more significant features, with the maximum value of the evaluation function. The subset of the selected features $X_{k}$ is obtained according to the following rule:

$$ X^{*} = X_{{k^{*} }} $$

$$ k^{*} = \mathop {\arg \hbox{Min} }\limits_{k} \{ J(X_{k} ) = J_{\hbox{Max} } \} $$

$$ J_{\hbox{Max} } = \mathop {\hbox{Max} }\limits_{k} \{ J(X_{k} )\} . $$

The selected evaluation function is the clustering accuracy of the FCM algorithm, which is described above.