In your diagram, I do not understand why you ignore the filter stages of a LPCM converter but include them with the DSD converter. If you ignore the low-pass filter stage of 1-bit DSD digital-to-analogue conversion, then a DSD-quantized sine wave would look jagged due to the presence of high-frequency ultrasonic quantization noise. The diagram below shows the noise introduced by a sigma-delta modulation ADC (as used in DSD) in quantizing a sine wave:
Page 31, 'Sigma Delta Modulation of A Chaotic Signal' Gary Ushaw PhD thesis, Edinburgh University.
http://www.era.lib.ed.ac.uk/retrieve/509/gu96.pdf
This quantization noise of a sine wave was generated in a simulation of a basic SDM ADC design. Different and more advanced SDM converter designs do exist, so the quantization noise shown here probably differs from that of a real-world SDM ADC.
DSD does not have a lower level of distortion than PCM. TPDF-dithered PCM has no distortion or modulation noise at all. Practical LPCM converters will still introduce errors of some kind. Please review Prof. Malcolm Hawksford's paper on LPCM versus DSD, Julian Dunn's description of dither, and this subjective double-blind study comparing LPCM and DSD:
Hawksford, M. (2001). "SDM versus LPCM: The Debate Continues", 110th AES Convention, paper 5397.
http://www.essex.ac.uk/ESE/research/audio_lab/malcolmspubdocs/C115 SDM versus LPCM.pdf
Dunn, J. (2003). "Measurement Techniques for Digital Audio", 'Dither Annex', p140-144. Audio Precision Application Note #5, Audio Precision, Inc. USA.
http://ap.com/library/technotes.htm
Blech, D. & Yang, M. (2004). "Perceptual Discrimination of Digital Coding Formats", Audio Engineering Society Convention Paper 6086, May 2004.
http://www.hfm-detmold.de/eti/projekte/diplomarbeiten/dsdvspcm/aes_paper_6086.pdf
A properly designed LPCM converter uses an analogue filter to remove unwanted high frequency/ultrasonic components from the output sine wave. With an oversampling LPCM converter, a digital and analogue filter would typically be used in combination.
The sinc function (the filters I mentioned approximate this function) allows perfect regeneration of the continuous form of the original bandlimited signal. This is the basis of the Nyquist-Shannon sampling theorem, which states that a signal can be regenerated with no loss of accuracy so long as the sample rate is twice that of the highest frequency components of the signal.
It is only necessary to use a higher sample rate if the filters of the PCM system are inadequate. This has not been demonstrated except in low quality designs, e.g., some computer sound cards, badly designed 'audiophile' DACs.