This paper presents experimental results on differentiating between healthy wheat plants and plants infected with Fusarium Head Blight (FHB) based on sensing the ambient gases in the plant environment using a gravimetric electronic nose enabled by a functionalized capacitive micromachined ultrasonic transducer (CMUT) array and machine learning (ML) algorithms. The CMUT sensor array is functionalized with organic/inorganic materials to capture disease-related volatile signals. The sensor data is processed and analyzed using ML algorithms for accurate plant classification. Experimental results demonstrate the effectiveness of the proposed approach in achieving high accuracy for plant disease detection at the end of the 11th day after plant inoculation.