Three-dimensional modelling of the middle-ear ossicular chain using a commercial high-resolution X-ray CT scanner

Abstract

The quantitative measurement of the three-dimensional (3-D) anatomy of the ear is of great importance in the making of teaching models and the design of mathematical models of parts of the ear, and also for the interpretation and presentation of experimental results. This article describes how we used virtual sections from a commercial high-resolution X-ray computed tomography (CT) scanner to make realistic 3-D anatomical models for various applications in our middle-ear research. The important problem of registration of the 3-D model within the experimental reference frame is discussed. The commercial X-ray CT apparatus is also compared with X-ray CT using synchrotron radiation, with magnetic resonance microscopy, with fluorescence optical sectioning, and with physical (histological) serial sections.

Figure 1

Schematic illustrating the design and working principles of the SkyScan 1072 X-ray micro-CT scanner. The object is placed on a rotating table in a cone beam of the X-ray source. The camera records X-ray images for a set of equally spaced rotational positions. The computer controls all settings of the equipment, the rotational stepping, and image recording. A “back projection” algorithm then calculates a stack of virtual serial sections, cross-sections at the level of each line in the X-ray images perpendicular to the imaging plane.

Figure 2

CT scan of an isolated stapes of a cat. a. A direct X-ray image of the stapes mounted upside-down in the scanner. The crura are almost parallel to the image plane. The dark band around the edge reveals how thin the bone is. b–f. Cross-sections of the stapes perpendicular to the axis running from the head to the footplate at the levels indicated by the horizontal lines [annotated (b)–(f)] in a. Some anatomical landmarks are annotated as CA: crus anterior, CP: crus posterior, H: head of stapes, FP: footplate, PSP: posterior stapedial process. (Please refer to text for more details).

Figure 3

Three-dimensional reconstructions of the stapes from six different viewing angles. a–d. The stapes is rotated in steps about the axis from footplate center to stapes head. e,f. Top and bottom views.

Figure 4

CT scan of an entire human temporal bone. a. Direct X-ray image of the temporal bone trimmed down into nearly a cylinder. b–f. Virtual sections at different heights of the specimen. The z levels where the sections are taken on the scale of a are annotated in the panels b–f. Some landmarks are annotated: anterior crus (CA), posterior crus (CP), head of the stapes (H), the bodies of the incus (I), and malleus (M), manubrium of the malleus (Man).

Figure 5

The 3-D reconstructions of the three middle-ear ossicles after segmentation of the serial section stack of the human temporal bone shown in Figure 4. Two views for each ossicle are shown. a,b. The malleus. c,d. The incus. e,f. The stapes. Viewing angles are different for each ossicle and chosen to highlight some details: For the malleus, the lateral process (LP) and the anterior process are indicated, on the incus the lenticular process (Len P), and for the stapes the anterior (CA) and posterior crus.

Figure 6

The total middle-ear chain reconstruction combining the ossicle models shown in Figure 5. The malleus, incus, and stapes are shown in their normal anatomical relation to each other. Different viewpoints are used to illustrate this relation better as explained in the text.

Figure 7

High-resolution 3-D models for the malleus, incus, and stapes based on separate scans of the isolated ossicles. The ossicles were dissected from the temporal bone that was used for the total temporal bone scan of Figure 5 and the corresponding reconstructions illustrated in Figure 6. Different viewpoints were chosen for the three ossicles.

Figure 8

The high-resolution 3-D model of the incus here is nicely replacing the low-resolution model in the total chain reconstruction shown in Figure 6. The position of the isolated incus model was calculated using an automatic registration procedure (ICP).

Figure 9

The registration of the 3-D model of the total middle-ear chain in the set of anatomical points for a cat ear. a. The model is roughly aligned with the anatomical data set (solid dots). b. We see how the ICP procedure has now aligned the model to fit the anatomical data points closely.