Introduction
Hans Chiari was the first who defined the term Chiari malformation (CM)
in 1891, specifying the malformation of varying degrees of the descent
of the brain stem and cerebellum below the foramen magnum (FM) (1). The
most common form of CM is type I, the elongation of the tonsils of the
cerebellum into the upper cervical canal. Although this situation is
usually presented as an incidental radiological finding, many symptoms
may also guide the practitioner to such a diagnosis (2).
The prevalent belief is that CM-I could result from an intrinsically
smaller overcrowded posterior fossa that predisposes the upper cervical
spine to eventual herniation of the cerebellar tonsils, resulting in
altered dynamics of Cerebrospinal fluid (CSF) flow (3). The known
primary radiological diagnosis of CM-I is relied on the degree of
tonsillar herniation (TH) below the FM. However, recent data also shows
the association of such malformation with smaller posterior cranial
fossa (PCF) volume and the anatomical issues regarding the Odontoid
(4-7). Recent studies suggest more radiological criteria that may aid to
the diagnosis of CM (5, 6, 7, 16, 18, 21).
Machine learning (ML) is one of the traditional methods of computer
systems that help to recognize and detect patterns in organizational and
commercial decisions (e.g., underlying dimensions/subgroups, nonlinear
associations/patterns, and so on) (8). This situation can be used in
differentiation, classifying, and organizing many health problems. ML
offers an opportunity to overcome obstacles using conventional
statistical methods.
This study presents the achieved results regarding some detected
potential radiological findings that may aid CM-I diagnosis using
several ML algorithms.