Preoperative Volume Determination for Pituitary Adenoma

📝 Abstract

The most common sellar lesion is the pituitary adenoma, and sellar tumors are approximately 10-15% of all intracranial neoplasms. Manual slice-by-slice segmentation takes quite some time that can be reduced by using the appropriate algorithms. In this contribution, we present a segmentation method for pituitary adenoma. The method is based on an algorithm that we have applied recently to segmenting glioblastoma multiforme. A modification of this scheme is used for adenoma segmentation that is much harder to perform, due to lack of contrast-enhanced boundaries. In our experimental evaluation, neurosurgeons performed manual slice-by-slice segmentation of ten magnetic resonance imaging (MRI) cases. The segmentations were compared to the segmentation results of the proposed method using the Dice Similarity Coefficient (DSC). The average DSC for all datasets was 75.92% +/- 7.24%. A manual segmentation took about four minutes and our algorithm required about one second.

💡 Analysis

The most common sellar lesion is the pituitary adenoma, and sellar tumors are approximately 10-15% of all intracranial neoplasms. Manual slice-by-slice segmentation takes quite some time that can be reduced by using the appropriate algorithms. In this contribution, we present a segmentation method for pituitary adenoma. The method is based on an algorithm that we have applied recently to segmenting glioblastoma multiforme. A modification of this scheme is used for adenoma segmentation that is much harder to perform, due to lack of contrast-enhanced boundaries. In our experimental evaluation, neurosurgeons performed manual slice-by-slice segmentation of ten magnetic resonance imaging (MRI) cases. The segmentations were compared to the segmentation results of the proposed method using the Dice Similarity Coefficient (DSC). The average DSC for all datasets was 75.92% +/- 7.24%. A manual segmentation took about four minutes and our algorithm required about one second.

📄 Content

Preoperative Volume Determination for Pituitary Adenoma Dženan Zukića *, Jan Eggerb, c, Miriam H. A. Bauerb, c, Daniela Kuhntb, Barbara Carlb Bernd Freislebenc, Andreas Kolba and Christopher Nimskyb a University of Siegen, Computer Graphics Group, Hölderlinstrasse 3, 57076 Siegen, Germany; b University of Marburg, Department of Neurosurgery, Baldingerstrasse, 35033 Marburg, Germany; c University of Marburg, Department of Mathematics and Computer Science, Hans-Meerwein-Str. 3, 35032 Marburg, Germany; ABSTRACT The most common sellar lesion is the pituitary adenoma, and sellar tumors are approximately 10-15% of all intracranial neoplasms. Manual slice-by-slice segmentation takes quite some time that can be reduced by using the appropriate algorithms. In this contribution, we present a segmentation method for pituitary adenoma. The method is based on an algorithm that we have applied recently to segmenting glioblastoma multiforme. A modification of this scheme is used for adenoma segmentation that is much harder to perform, due to lack of contrast-enhanced boundaries. In our experimental evaluation, neurosurgeons performed manual slice-by-slice segmentation of ten magnetic resonance imaging (MRI) cases. The segmentations were compared to the segmentation results of the proposed method using the Dice Similarity Coefficient (DSC). The average DSC for all datasets was 75.92%±7.24%. A manual segmentation took about four minutes and our algorithm required about one second. Keywords: Pituitary Adenoma, Preoperative, Volume Determination, MRI, Balloon Inflation

1. INTRODUCTION Approximately 10-15% of all intracranial neoplasms are sellar tumors. The most common sellar lesion is the pituitary adenoma1,2. The lesions can be classified according to size or hormone-secretion (hormone-active and hormone- inactive). Microadenomas are less than 1 cm in diameter, whereas macroadenomas measure more than 1 cm. The rare giant-adenomas have more than 4 cm in diameter. Secreted hormones can be cortisol (Cushing’s disease), human growth hormone (hGH; acromegaly), follicle- stimulating hormone (FSH), luteinising hormone (LH), thyroid-stimulating hormone (TSH), prolactine, or a combination of these. Only for the prolactine-expressing tumors, a pharmacological treatment is the initial treatment of choice in form of dopamine-agonists. Treatment is most commonly followed by a decrease of prolactine-levels and tumor volume. For acromegaly and Cushing’s disease, surgery remains the first-line treatment, although somatostatin receptor analogues or combined dopamine/somatostatin receptor analogues are a useful second-line therapeutical option for hGH-expressing tumors. Current medical therapies for Cushing’s disease primarily focus on the adrenal blockade of cortisol production, although pasireotide and cabergoline show promise as pituitary-directed medical therapy for Cushing’s disease. Thus, not only for the most hormone-active, but also for hormone-inactive macroadenomas with mass-effect, surgery is the treatment of choice, most possibly via a transsphenoidal approach3. For hormone-inactive mircroadenomas (<1cm) there is no need for immediate surgical resection. The follow-up contains endocrine and ophthalmological evaluation as well as magnetic resonance imaging (MRI). In case of continuous tumor volume progress, microsurgical excision is the treatment of choice. Thus, the tumor volume should be tracked over the time of the follow-up. In this contribution, we present a segmentation method for pituitary adenomas. The method is based on an algorithm we recently developed for segmenting glioblastoma multiforme (GBM)4. The paper is organized as follows. Section 2 discusses related work. Section 3 presents the details of the proposed approach. Section 4 discusses experimental results and concludes the paper.

• Further author information: (Send correspondence to Dž. Z.) Dž. Z.: E-mail: zukic@fb12.uni-siegen.de, Telephone: +49 271 740 2826 J. E.: E-mail: egger@med.uni-marburg.de, Telephone: +49 6421 58 66754

2. RELATED WORK Several algorithms have already been proposed for segmenting brain tumors in MRI (magnetic resonance images). A relatively recent overview of some deterministic and statistical approaches is given by Angelini et al.5. Most of these approaches are region-based; more recent ones are based on deformable models and include edge-information. Neubauer et al.6 and Wolfsberger et al.7 introduce STEPS, a virtual endoscopy system designed to aid surgery of pituitary tumors. STEPS uses a semi-automatic segmentation method that is based on the so-called watershed-from- markers technique. The watershed-from-markers technique uses manually defined markers in the object of interest and the background. A memory efficient and fast implementation of the watershed-from-markers algorithm – also extended to 3D – has been developed by Felk

This content is AI-processed based on ArXiv data.