Threedimensional ultrasound for the assessment of breast lesions

1
Three-dimensional ultrasound for the assessment
of breast lesions

• D. O. WATERMANN, M. FO LDI, A. HANJALIC-BECK, A.
  HASENBURG, A. LU GHAUSEN, H. PRO MPELER, G.
  GITSCH and E. STICKELER
• Freiburg University Medical Center, Obstetrics
  and Gynecology, Freiburg, Germany
• Ultrasound Obstet Gynecol 2005

Reporter Chia-Chia Hsu
2
Outline

• Introduction
• Methods
• Results
• Conclusions

3
Introduction

• To evaluate the diagnostic accuracy of three
  dimensional (3D) ultrasound in comparison with
  conventional two-dimensional (2D) ultrasound in
  the characterization of breast lesions.

4
Methods

• Patients
• 100 breast lesions (57 malignant, 43 benign)

5
Imaging analysis

• Ultrasound criteria for analysis (Table 2)
• 2D US 10 criteria
• 3D US 13 criteria

6
(No Transcript)
7
Ultrasound criteria (Sagittal)

• Shape
• Orientation
• Echogenicity
• Echo pattern
• Hyperechogenic spots
• Margin contour
• Margin
• Lesion boundary
• Posterior acoustic features
• Surrounding tissue

8
Ultrasound criteria (Coronal)

• Retraction
• Margin contour
• Margin

9
(No Transcript)
10
(No Transcript)
11
Imaging analysis (cont.)

• BI-RADS (Breast Imaging Reporting and Data
  System)
• Category 1 normal breast imaging without finding
• Category 2 lesion very likely to be benign
• Category 3 lesion unlikely to be malignant
• Category 4 suspicious lesions with a
  moderate-to-high probability of being malignant
• Category 5 reserved for tumors
• Benign class (categories 2 and 3)
• Malignant class (categories 4 and 5)

12
Data management and statistics

• Spearmans rank correlation coefficients (r)
• To assess the relationship between the different
  ultrasound characteristics.
• kappa (?)
• To compare the relationship between 2D and 3D
  ultrasound Characteristics.
• The ability of ultrasound criteria to determine
  the histological character of a lesion was
  analyzed by the Chi-square test and expressed as
  odds ratio (OR) with 95 CI.
• P-value
• P lt 0.05 was considered statistically
  significant.
• Performed binary logistic-regression analyses
  with the investigated ultrasound criteria as
  independent variables.
• ROC curves
• were constructed to obtain an index for the
  diagnostic accuracy of 2D and 3D ultrasound.

13
Results

• 6 investigators analyzed independently 2D and 3D
  ultrasound images of 100 breast lesions.
• 10 2D and 13 3D ultrasound criteria were
  determined.
• The lesions categorized into 4 classes according
  to a modified BI-RADS classification system.

14
Results (cont.)

• Correlation between ultrasound criteria
• The relationship between the investigated
  ultrasound criteria is shown in Table 3.
• Comparison between 2D and 3D ultrasound criteria.
• criteria indicating malignancy were more often
  described in 3D than in 2D ultrasound.

15
(No Transcript)
16
Evaluation of 2D and 3D ultrasound criteria for
lesion characterization

• Most of the criteria were highly significant in
  characterizing breast lesions (P 0.001).
• Only hyperechogenic spots were not significant
  and did not contribute to lesion characterization
  in 2D or 3D ultrasound.
• In 2D
• surrounding tissue, shape, margin contour, lesion
  boundary and posterior acoustic features
• In 3D
• the retraction criterion in the coronal plane
• the orientation, margin, margin contour and
  surrounding tissue criteria in the sagittal plane

17
(No Transcript)
18
Value of the coronal plane in 3D ultrasound

• The coronal plane is unique to the 3D technique.
• The retraction phenomenon
• Expression of margin contour and margin in the
  coronal plane.
• Be independent and significant factors.
• Had a high specificity(94.6) but a low
  sensitivity (54.7), with an accuracy to
  determine malignant and benign lesions of 71.8.
• The false-positive rate was 7.0 and the
  false-negative rate 38.8.

19
Value of the coronal plane in 3D ultrasound

• The indistinct margin feature
• Did not significantly differ between the coronal
  plane and conventional planes.
• An irregular or spiculate margin contour
• was more frequently observed in the coronal plane
  than in the conventional planes.
• This feature had a high sensitivity (95.0), a
  low specificity (46.9) and an accuracy of 66.2
  to classify a malignant breast lesion.

20
Comparison of the diagnostic value of 2D and 3D
ultrasound

• 2D 0.846
• 3D 0.851

21
Conclusions

• Ultrasound features on 3D US differ significantly
  from those on 2D US.
• The diagnostic accuracy of both methods is almost
  identical.
• 3D US as an adjunct to conventional 2D US should
  be evaluated in larger trials to determine its
  clinical value in breast imaging.